From 90f0257ae0f3d3c5a319d95555f4f7a0a42d0d85 Mon Sep 17 00:00:00 2001 From: SSKJ Dev Date: Fri, 17 Jul 2026 02:19:22 +0000 Subject: [PATCH] update result --- .../adaptive_points.jsonl | 132 ++++++++++++++++++ .../adaptive_shapes.jsonl | 44 ++++++ .../adaptive_summary.jsonl | 71 ++++++++++ .../adaptive_summary.md | 77 ++++++++++ .../metrics/vllm_c112_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i131072_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o2048_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o4096_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o512_a1.json | 25 ++++ .../metrics/vllm_c1_i131072_o128_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o1024_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o128_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o256_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o512_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o128_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c32_i16384_o1024_a1.json | 25 ++++ .../metrics/vllm_c32_i16384_o2048_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o1024_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o256_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o4096_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o512_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c48_i4096_o1024_a1.json | 25 ++++ .../metrics/vllm_c48_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c48_i4096_o512_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c64_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c80_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c80_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c80_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c80_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c8_i1024_o256_a1.json | 25 ++++ .../metrics/vllm_c8_i131072_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i16384_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i16384_o256_a1.json | 25 ++++ .../metrics/vllm_c8_i16384_o512_a1.json | 25 ++++ .../metrics/vllm_c8_i4096_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o1024_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o256_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o512_a1.json | 25 ++++ .../metrics/vllm_c96_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c96_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c96_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c16_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i131072_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o2048_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i16384_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o4096_a1.json | 25 ++++ .../metrics/vllm_c16_i4096_o512_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o1024_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o128_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o256_a1.json | 25 ++++ .../metrics/vllm_c16_i65536_o512_a1.json | 25 ++++ .../metrics/vllm_c1_i1024_o128_a1.json | 25 ++++ .../metrics/vllm_c1_i1024_o256_a1.json | 25 ++++ .../metrics/vllm_c1_i131072_o128_a1.json | 25 ++++ .../metrics/vllm_c1_i16384_o512_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o1024_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o128_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o256_a1.json | 25 ++++ .../metrics/vllm_c1_i65536_o512_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c32_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c32_i16384_o2048_a1.json | 25 ++++ .../metrics/vllm_c32_i16384_o256_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o1024_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o256_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o4096_a1.json | 25 ++++ .../metrics/vllm_c32_i4096_o512_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c48_i1024_o512_a1.json | 25 ++++ .../metrics/vllm_c48_i4096_o2048_a1.json | 25 ++++ .../metrics/vllm_c48_i4096_o512_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o2048_a1.json | 25 ++++ .../metrics/vllm_c64_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c80_i1024_o1024_a1.json | 25 ++++ .../metrics/vllm_c80_i1024_o4096_a1.json | 25 ++++ .../metrics/vllm_c8_i1024_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i1024_o256_a1.json | 25 ++++ .../metrics/vllm_c8_i131072_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i16384_o1024_a1.json | 25 ++++ .../metrics/vllm_c8_i16384_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i16384_o512_a1.json | 25 ++++ .../metrics/vllm_c8_i4096_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o1024_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o128_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o256_a1.json | 25 ++++ .../metrics/vllm_c8_i65536_o512_a1.json | 25 ++++ .../tp8_dp1/server_cmd.txt | 1 + 137 files changed, 3625 insertions(+) create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.jsonl create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.md create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c112_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i131072_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i131072_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i1024_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i131072_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i4096_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i131072_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i131072_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i16384_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o2048_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o4096_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i131072_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i4096_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o1024_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o128_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o256_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o512_a1.json create mode 100644 experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/server_cmd.txt diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_points.jsonl b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_points.jsonl index d15141e..d4913ae 100644 --- a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_points.jsonl +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_points.jsonl @@ -58,3 +58,135 @@ {"status":"COMPLETED","completed":40,"failed":0,"duration_s":818.7336148600007,"request_tps":0.04885593955591991,"input_tps":12807.291418947068,"output_tps":12.507120526315497,"total_tps":12819.798539473382,"mean_input_tokens":262144,"mean_output_tokens":256,"ttft_p50_ms":105032.74306450476,"ttft_p95_ms":162785.75602490891,"ttft_p99_ms":165083.89961342298,"tpot_p50_ms":214.14210347058685,"tpot_p95_ms":422.65357399257476,"tpot_p99_ms":440.83721461195177,"e2e_p50_ms":164185.14075000712,"e2e_p95_ms":168861.28058029353,"e2e_p99_ms":169246.668313704,"itl_p50_ms":13.376075992709957,"itl_p95_ms":1720.7928034986253,"itl_p99_ms":2045.3944211971263,"validation_errors":[],"timestamp":"2026-07-16T11:23:47+00:00","engine":"vllm","tp":2,"dp":4,"mark":"Y","isl":262144,"osl":256,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp2_dp4/raw_outputs/vllm_adaptive_c8_i262144_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp2_dp4/logs/vllm_c8_i262144_o256_a1.log"} {"status":"COMPLETED","completed":5,"failed":0,"duration_s":222.6894240649999,"request_tps":0.022452795057481358,"input_tps":5885.865507548393,"output_tps":5.7479155347152275,"total_tps":5891.6134230831085,"mean_input_tokens":262144,"mean_output_tokens":256,"ttft_p50_ms":41992.54209799983,"ttft_p95_ms":42385.26861999417,"ttft_p99_ms":42405.3075527912,"tpot_p50_ms":9.564196976419884,"tpot_p95_ms":9.778543265853717,"tpot_p99_ms":9.797764013144736,"e2e_p50_ms":44431.4123269869,"e2e_p95_ms":44819.5239008026,"e2e_p99_ms":44832.609473761986,"itl_p50_ms":10.17798750399379,"itl_p95_ms":10.34693365872954,"itl_p99_ms":10.991512061009418,"validation_errors":[],"timestamp":"2026-07-16T11:28:54+00:00","engine":"vllm","tp":2,"dp":4,"mark":"Y","isl":262144,"osl":256,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp2_dp4/raw_outputs/vllm_adaptive_c1_i262144_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp2_dp4/logs/vllm_c1_i262144_o256_a1.log"} {"timestamp":"2026-07-16T12:39:42+00:00","engine":"vllm","tp":2,"dp":4,"mark":"Y","isl":524288,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":2,"status":"FAILED","completed":0,"failed":80,"error_type":"benchmark rc=124","validation_errors":[],"gain_pct":null,"plateau_streak":0,"raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp2_dp4/raw_outputs/vllm_adaptive_c16_i524288_o128_a2.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp2_dp4/logs/vllm_c16_i524288_o128_a2.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":52.56799242499983,"request_tps":1.5218386000594972,"input_tps":1558.3627264609252,"output_tps":194.79534080761564,"total_tps":1753.158067268541,"mean_input_tokens":1024.0,"mean_output_tokens":128.0,"ttft_p50_ms":937.427844000922,"ttft_p95_ms":1428.1466051499592,"ttft_p99_ms":18393.34938969926,"tpot_p50_ms":19.7043993346339,"tpot_p95_ms":181.49350103310894,"tpot_p99_ms":182.55337019226835,"e2e_p50_ms":3306.429382006172,"e2e_p95_ms":23952.145700105757,"e2e_p99_ms":23955.10612057842,"itl_p50_ms":14.576606998161878,"itl_p95_ms":15.284183846233645,"itl_p99_ms":20.36303154163767,"validation_errors":[],"timestamp":"2026-07-16T13:12:16+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i1024_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i1024_o128_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":95.59839020299842,"request_tps":1.6736683500658116,"input_tps":1713.836390467391,"output_tps":214.22954880842389,"total_tps":1928.0659392758148,"mean_input_tokens":1024.0,"mean_output_tokens":128.0,"ttft_p50_ms":4119.889060501009,"ttft_p95_ms":26025.381604798167,"ttft_p99_ms":26478.019099681696,"tpot_p50_ms":21.413118271642976,"tpot_p95_ms":376.97693987998275,"tpot_p99_ms":387.07957727625,"e2e_p50_ms":7732.499119003478,"e2e_p95_ms":51129.2892126934,"e2e_p99_ms":51135.8539534836,"itl_p50_ms":16.923131996009033,"itl_p95_ms":18.388393107306907,"itl_p99_ms":244.194259904325,"validation_errors":[],"timestamp":"2026-07-16T13:14:35+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":128,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":9.976731,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i1024_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i1024_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":48.607205292995786,"request_tps":1.6458465266162476,"input_tps":1685.3468432550376,"output_tps":421.3367108137594,"total_tps":2106.683554068797,"mean_input_tokens":1024.0,"mean_output_tokens":256.0,"ttft_p50_ms":782.0743444972322,"ttft_p95_ms":12196.891359947767,"ttft_p99_ms":12266.554353885149,"tpot_p50_ms":15.961954474515354,"tpot_p95_ms":58.48537955195449,"tpot_p99_ms":60.24822486883116,"e2e_p50_ms":8633.398842990573,"e2e_p95_ms":16283.070811999643,"e2e_p99_ms":16288.124511987555,"itl_p50_ms":14.68896050209878,"itl_p95_ms":15.627986006438732,"itl_p99_ms":16.497610333171902,"validation_errors":[],"timestamp":"2026-07-16T13:16:07+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i1024_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i1024_o256_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":18.495949040006963,"request_tps":2.1626357162576255,"input_tps":2214.5389734478085,"output_tps":553.6347433619521,"total_tps":2768.1737168097607,"mean_input_tokens":1024.0,"mean_output_tokens":256.0,"ttft_p50_ms":440.5348879954545,"ttft_p95_ms":3502.821924509044,"ttft_p99_ms":3505.5937448414625,"tpot_p50_ms":10.14652707840558,"tpot_p95_ms":22.57474098336602,"tpot_p99_ms":23.09809012723831,"e2e_p50_ms":3156.6136774999904,"e2e_p95_ms":6075.7029040003545,"e2e_p99_ms":6078.332349029661,"itl_p50_ms":10.086970491101965,"itl_p95_ms":10.408660549001068,"itl_p99_ms":11.378483117441656,"validation_errors":[],"timestamp":"2026-07-16T13:17:38+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":256,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i1024_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i1024_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":37.51114835499902,"request_tps":2.132699304294655,"input_tps":2183.8840875977266,"output_tps":1091.9420437988633,"total_tps":3275.82613139659,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":565.4102255066391,"ttft_p95_ms":804.2641735068173,"ttft_p99_ms":840.2388689920183,"tpot_p50_ms":12.695262536184165,"tpot_p95_ms":16.153524499127695,"tpot_p99_ms":16.279702295489752,"e2e_p50_ms":6936.251829996763,"e2e_p95_ms":8593.082947545918,"e2e_p99_ms":8608.418859792146,"itl_p50_ms":12.14090500434395,"itl_p95_ms":15.356732595682843,"itl_p99_ms":15.861063279880907,"validation_errors":[],"timestamp":"2026-07-16T13:18:52+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":44.825771888004965,"request_tps":3.569375233509694,"input_tps":3655.040239113927,"output_tps":1827.5201195569634,"total_tps":5482.56035867089,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":912.5870395073434,"ttft_p95_ms":1430.6775802040647,"ttft_p99_ms":1454.0497396518183,"tpot_p50_ms":15.280834627210176,"tpot_p95_ms":18.225302902624872,"tpot_p99_ms":18.700063700290844,"e2e_p50_ms":8596.120701506152,"e2e_p95_ms":10176.577183793415,"e2e_p99_ms":10183.192389139585,"itl_p50_ms":14.963374997023493,"itl_p95_ms":17.358059901744127,"itl_p99_ms":18.431273359310598,"validation_errors":[],"timestamp":"2026-07-16T13:20:15+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":67.364205,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":57.3580158239929,"request_tps":4.184245158278432,"input_tps":4284.667042077114,"output_tps":2142.333521038557,"total_tps":6427.000563115672,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":1114.938983497268,"ttft_p95_ms":2623.5857654435677,"ttft_p99_ms":2657.5968347305025,"tpot_p50_ms":19.735206197671033,"tpot_p95_ms":22.902151353534165,"tpot_p99_ms":23.980957457851613,"e2e_p50_ms":11155.085417994997,"e2e_p95_ms":12560.576499555464,"e2e_p99_ms":12733.288235523505,"itl_p50_ms":19.239879009546712,"itl_p95_ms":20.936376793542877,"itl_p99_ms":22.45631740195677,"validation_errors":[],"timestamp":"2026-07-16T13:21:55+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":17.226262,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":64.12715022399789,"request_tps":4.990086084945788,"input_tps":5109.848150984487,"output_tps":2554.9240754922434,"total_tps":7664.77222647673,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":1574.427007995837,"ttft_p95_ms":2990.9389156964608,"ttft_p99_ms":3418.0432158199255,"tpot_p50_ms":21.48998137866181,"tpot_p95_ms":26.06068395949076,"tpot_p99_ms":26.0992564425009,"e2e_p50_ms":13091.199074500764,"e2e_p95_ms":14394.905802008725,"e2e_p99_ms":14404.70782280463,"itl_p50_ms":20.648846999392845,"itl_p95_ms":22.798411999247037,"itl_p99_ms":25.05893900524825,"validation_errors":[],"timestamp":"2026-07-16T13:23:41+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":19.258932,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c64_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c64_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":400,"failed":0,"duration_s":69.2822109109984,"request_tps":5.773487807914353,"input_tps":5912.051515304297,"output_tps":2956.0257576521485,"total_tps":8868.077272956445,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":1404.177324497141,"ttft_p95_ms":3550.293400152441,"ttft_p99_ms":4256.765821308945,"tpot_p50_ms":23.07539692855886,"tpot_p95_ms":26.923700127783786,"tpot_p99_ms":28.125053756603382,"e2e_p50_ms":13326.003155001672,"e2e_p95_ms":14892.070211340615,"e2e_p99_ms":14897.991692002543,"itl_p50_ms":22.541024998645298,"itl_p95_ms":24.51956669683568,"itl_p99_ms":27.620439298043483,"validation_errors":[],"timestamp":"2026-07-16T13:25:34+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"concurrency":80,"num_prompts":400,"warmup_requests":80,"attempt":1,"gain_pct":15.699162,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c80_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c80_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":480,"failed":0,"duration_s":75.0480901490082,"request_tps":6.3958989368944446,"input_tps":6549.400511379911,"output_tps":3274.7002556899556,"total_tps":9824.100767069867,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":1664.6413179987576,"ttft_p95_ms":4356.379911445401,"ttft_p99_ms":4993.593463639991,"tpot_p50_ms":25.16758154697011,"tpot_p95_ms":30.006507990595942,"tpot_p99_ms":31.581847599314614,"e2e_p50_ms":14630.724068490963,"e2e_p95_ms":17554.996255355218,"e2e_p99_ms":17564.006839842623,"itl_p50_ms":23.962745995959267,"itl_p95_ms":26.025201004813425,"itl_p99_ms":29.676723000011375,"validation_errors":[],"timestamp":"2026-07-16T13:27:33+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"concurrency":96,"num_prompts":480,"warmup_requests":96,"attempt":1,"gain_pct":10.780505,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c96_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c96_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":70.41322445700644,"request_tps":1.1361502134992716,"input_tps":1163.417818623254,"output_tps":1163.417818623254,"total_tps":2326.835637246508,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":421.1962485060212,"ttft_p95_ms":677.2944091026147,"ttft_p99_ms":691.2584483872342,"tpot_p50_ms":12.164721241445186,"tpot_p95_ms":15.51125149233277,"tpot_p99_ms":15.721278570422475,"e2e_p50_ms":12909.268272502231,"e2e_p95_ms":16315.028272505877,"e2e_p99_ms":16321.569995415775,"itl_p50_ms":12.13237400224898,"itl_p95_ms":15.468433196656404,"itl_p99_ms":16.074650280061178,"validation_errors":[],"timestamp":"2026-07-16T13:29:25+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":85.3312213080062,"request_tps":1.875046407955115,"input_tps":1920.0475217460378,"output_tps":1920.0475217460378,"total_tps":3840.0950434920755,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":806.487787005608,"ttft_p95_ms":900.6144404585939,"ttft_p99_ms":904.4747244215978,"tpot_p50_ms":15.241720033234294,"tpot_p95_ms":17.703688128979007,"tpot_p99_ms":17.972658989592805,"e2e_p50_ms":16181.125543495,"e2e_p95_ms":18731.30079265975,"e2e_p99_ms":18746.509167428594,"itl_p50_ms":15.200200999970548,"itl_p95_ms":17.96690479677636,"itl_p99_ms":18.968004560447298,"validation_errors":[],"timestamp":"2026-07-16T13:31:34+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":65.035079,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":97.32574620700325,"request_tps":2.46594564494313,"input_tps":2525.128340421765,"output_tps":2525.128340421765,"total_tps":5050.25668084353,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":971.1724649969256,"ttft_p95_ms":1047.563062050176,"ttft_p99_ms":1059.1346113267355,"tpot_p50_ms":17.463300068922933,"tpot_p95_ms":20.0403643846501,"tpot_p99_ms":20.23941422052442,"e2e_p50_ms":18865.876165502414,"e2e_p95_ms":20948.537042703538,"e2e_p99_ms":20966.204005898326,"itl_p50_ms":17.521906993351877,"itl_p95_ms":19.884526249370538,"itl_p99_ms":20.897341451927787,"validation_errors":[],"timestamp":"2026-07-16T13:33:57+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":31.513846,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":111.44104069900641,"request_tps":2.8714735432549947,"input_tps":2940.3889082931146,"output_tps":2940.3889082931146,"total_tps":5880.777816586229,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1251.2050785007887,"ttft_p95_ms":1664.9364830955165,"ttft_p99_ms":1687.642185555742,"tpot_p50_ms":20.892368739980142,"tpot_p95_ms":22.107194719502488,"tpot_p99_ms":22.406787545171895,"e2e_p50_ms":22625.58904299658,"e2e_p95_ms":23358.138840601896,"e2e_p99_ms":23385.716578069405,"itl_p50_ms":20.738980994792655,"itl_p95_ms":21.87385440629441,"itl_p99_ms":24.061159000266343,"validation_errors":[],"timestamp":"2026-07-16T13:36:35+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":16.445127,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c64_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c64_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":400,"failed":0,"duration_s":123.47737352400145,"request_tps":3.2394598992846917,"input_tps":3317.2069368675243,"output_tps":3317.2069368675243,"total_tps":6634.413873735049,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1259.8191590004717,"ttft_p95_ms":2176.3979704985104,"ttft_p99_ms":2561.1882165598217,"tpot_p50_ms":23.053262887589092,"tpot_p95_ms":23.774626320429117,"tpot_p99_ms":24.000006763245814,"e2e_p50_ms":24829.307881504064,"e2e_p95_ms":24994.919799604395,"e2e_p99_ms":25020.736574648326,"itl_p50_ms":22.634291992289945,"itl_p95_ms":24.973727005999535,"itl_p99_ms":27.328714399482124,"validation_errors":[],"timestamp":"2026-07-16T13:39:20+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":80,"num_prompts":400,"warmup_requests":80,"attempt":1,"gain_pct":12.815245,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c80_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c80_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":480,"failed":0,"duration_s":131.60086494100688,"request_tps":3.647392440886852,"input_tps":3734.9298594681363,"output_tps":3734.9298594681363,"total_tps":7469.859718936273,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1641.8105130069307,"ttft_p95_ms":2725.6738467483956,"ttft_p99_ms":2848.544350732118,"tpot_p50_ms":24.315747637830274,"tpot_p95_ms":25.91790127140095,"tpot_p99_ms":26.00329389268108,"e2e_p50_ms":26516.128896502778,"e2e_p95_ms":27687.729706803657,"e2e_p99_ms":27709.92693845037,"itl_p50_ms":23.818820001906715,"itl_p95_ms":26.684747799299657,"itl_p99_ms":30.725880123209194,"validation_errors":[],"timestamp":"2026-07-16T13:42:18+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":96,"num_prompts":480,"warmup_requests":96,"attempt":1,"gain_pct":12.59261,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c96_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c96_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":560,"failed":0,"duration_s":143.62631484199665,"request_tps":3.899006951588615,"input_tps":3992.5831184267417,"output_tps":3992.5831184267417,"total_tps":7985.1662368534835,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1571.4164454911952,"ttft_p95_ms":4123.8772861033785,"ttft_p99_ms":4936.747990879667,"tpot_p50_ms":25.948950879762915,"tpot_p95_ms":28.919239563878318,"tpot_p99_ms":29.696389907960754,"e2e_p50_ms":28112.99813849473,"e2e_p95_ms":31451.82638790211,"e2e_p99_ms":31470.12428371294,"itl_p50_ms":25.19894449505955,"itl_p95_ms":28.669426793931052,"itl_p99_ms":31.52743758109864,"validation_errors":[],"timestamp":"2026-07-16T13:45:29+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"concurrency":112,"num_prompts":560,"warmup_requests":112,"attempt":1,"gain_pct":6.898477,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c112_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c112_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":137.77100664599857,"request_tps":0.5806736986800084,"input_tps":594.6098674483286,"output_tps":1189.2197348966572,"total_tps":1783.8296023449857,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":509.03518099948997,"ttft_p95_ms":817.3343080510676,"ttft_p99_ms":821.0908090611338,"tpot_p50_ms":12.053855090129739,"tpot_p95_ms":15.30965305794184,"tpot_p99_ms":15.332859359112767,"e2e_p50_ms":25319.949440003256,"e2e_p95_ms":31725.24674650267,"e2e_p99_ms":31736.754303508642,"itl_p50_ms":12.262283504242077,"itl_p95_ms":15.428829462325668,"itl_p99_ms":15.964222954062278,"validation_errors":[],"timestamp":"2026-07-16T13:48:26+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":2048,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":162.71412774299097,"request_tps":0.9833196552712499,"input_tps":1006.9193269977599,"output_tps":2013.8386539955197,"total_tps":3020.75798099328,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":871.0656835028203,"ttft_p95_ms":1184.4237971468829,"ttft_p99_ms":1192.6747960213106,"tpot_p50_ms":15.119994677330888,"tpot_p95_ms":17.311258725843935,"tpot_p99_ms":17.470737866078203,"e2e_p50_ms":31956.353274501453,"e2e_p95_ms":36126.69280984701,"e2e_p99_ms":36139.65432776953,"itl_p50_ms":14.982249005697668,"itl_p95_ms":17.40834979864303,"itl_p99_ms":18.12914603331592,"validation_errors":[],"timestamp":"2026-07-16T13:51:52+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":2048,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":69.341173,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":205.12268571999448,"request_tps":1.170031482171676,"input_tps":1198.1122377437962,"output_tps":2396.2244754875924,"total_tps":3594.3367132313883,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1103.477151504194,"ttft_p95_ms":1655.3460868577529,"ttft_p99_ms":2063.1610584430746,"tpot_p50_ms":19.476409309720992,"tpot_p95_ms":19.98451202838762,"tpot_p99_ms":20.224875959688827,"e2e_p50_ms":41019.6206035107,"e2e_p95_ms":41983.93201486106,"e2e_p99_ms":42004.45509949772,"itl_p50_ms":19.33959749294445,"itl_p95_ms":21.063311200850876,"itl_p99_ms":22.709367697680136,"validation_errors":[],"timestamp":"2026-07-16T13:56:02+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":2048,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":18.987908,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":227.52200377998815,"request_tps":1.4064573741598956,"input_tps":1440.212351139733,"output_tps":2880.424702279466,"total_tps":4320.637053419199,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1391.4943349955138,"ttft_p95_ms":2766.4422078472853,"ttft_p99_ms":3516.292367083952,"tpot_p50_ms":21.495228763068177,"tpot_p95_ms":22.537871907642682,"tpot_p99_ms":22.736073239234297,"e2e_p50_ms":45078.62572550221,"e2e_p95_ms":47292.98801284676,"e2e_p99_ms":47307.55221481537,"itl_p50_ms":21.152492990950122,"itl_p95_ms":22.822722894488834,"itl_p99_ms":24.930158918141395,"validation_errors":[],"timestamp":"2026-07-16T14:00:36+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":2048,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":20.206797,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c64_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c64_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":400,"failed":0,"duration_s":265.8902801889926,"request_tps":1.5043799258689836,"input_tps":1540.4850440898392,"output_tps":3080.9700881796784,"total_tps":4621.455132269518,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1453.8656489967252,"ttft_p95_ms":4220.62352244684,"ttft_p99_ms":4392.5506810807565,"tpot_p50_ms":23.16868536492179,"tpot_p95_ms":32.49832417950609,"tpot_p99_ms":32.50333228816742,"e2e_p50_ms":49589.71686950099,"e2e_p95_ms":67762.65414134323,"e2e_p99_ms":67792.98369047901,"itl_p50_ms":22.87503899424337,"itl_p95_ms":24.68757720780559,"itl_p99_ms":27.605341025628153,"validation_errors":[],"timestamp":"2026-07-16T14:05:45+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":2048,"concurrency":80,"num_prompts":400,"warmup_requests":80,"attempt":1,"gain_pct":6.962355,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c80_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c80_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":271.16081980899617,"request_tps":0.29502787333491415,"input_tps":302.1085422949521,"output_tps":1208.4341691798084,"total_tps":1510.5427114747604,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":627.1758095026598,"ttft_p95_ms":960.0995247899846,"ttft_p99_ms":971.8106476443063,"tpot_p50_ms":11.913466125030766,"tpot_p95_ms":15.119212534556272,"tpot_p99_ms":15.134965707044527,"e2e_p50_ms":49374.50365599943,"e2e_p95_ms":62329.371708749386,"e2e_p99_ms":62340.975158869114,"itl_p50_ms":12.052630001562648,"itl_p95_ms":15.1747610012535,"itl_p99_ms":15.672232958022514,"validation_errors":[],"timestamp":"2026-07-16T14:11:02+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":318.9912624319986,"request_tps":0.50158113667489,"input_tps":513.6190839550874,"output_tps":2054.4763358203495,"total_tps":2568.0954197754368,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":867.0998999950825,"ttft_p95_ms":1568.6099555969122,"ttft_p99_ms":1761.4919271026156,"tpot_p50_ms":14.940362645543985,"tpot_p95_ms":17.25645533934084,"tpot_p99_ms":17.34749941946349,"e2e_p50_ms":62241.984514999785,"e2e_p95_ms":71418.52431923835,"e2e_p99_ms":71433.78462005581,"itl_p50_ms":14.883934498357121,"itl_p95_ms":17.468459051451646,"itl_p99_ms":18.707020131550955,"validation_errors":[],"timestamp":"2026-07-16T14:17:03+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":70.01144,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":387.6589750939893,"request_tps":0.6191008474441,"input_tps":633.9592677827584,"output_tps":2535.8370711310336,"total_tps":3169.7963389137917,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1218.414176008082,"ttft_p95_ms":2534.173024399205,"ttft_p99_ms":2783.003787570633,"tpot_p50_ms":19.216241195482507,"tpot_p95_ms":19.885012049902812,"tpot_p99_ms":19.88576329422029,"e2e_p50_ms":79752.23233650468,"e2e_p95_ms":82442.38089720166,"e2e_p99_ms":82460.91719069882,"itl_p50_ms":18.868017999920994,"itl_p95_ms":20.22105274882051,"itl_p99_ms":21.84910236479482,"validation_errors":[],"timestamp":"2026-07-16T14:24:14+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":23.429851,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":429.61134696399677,"request_tps":0.7448592833066333,"input_tps":762.7359061059925,"output_tps":3050.94362442397,"total_tps":3813.6795305299624,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1419.7698949938058,"ttft_p95_ms":2921.583319955971,"ttft_p99_ms":3675.6699006249255,"tpot_p50_ms":20.945350810012243,"tpot_p95_ms":21.617453843995357,"tpot_p99_ms":21.769692606291244,"e2e_p50_ms":87146.0334000003,"e2e_p95_ms":89824.40359709944,"e2e_p99_ms":89989.40303418975,"itl_p50_ms":20.697043000836857,"itl_p95_ms":22.144093887618506,"itl_p99_ms":24.082228924962692,"validation_errors":[],"timestamp":"2026-07-16T14:32:08+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":20.313078,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c64_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c64_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":400,"failed":0,"duration_s":461.13784259100794,"request_tps":0.8674195935699159,"input_tps":888.2376638155939,"output_tps":3552.9506552623757,"total_tps":4441.18831907797,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1903.3365680006682,"ttft_p95_ms":3811.413884991634,"ttft_p99_ms":4251.1653185045,"tpot_p50_ms":21.528175279001097,"tpot_p95_ms":23.315381058901853,"tpot_p99_ms":23.403718928096588,"e2e_p50_ms":90469.03869150265,"e2e_p95_ms":96639.56603370316,"e2e_p99_ms":96666.38061486068,"itl_p50_ms":21.778908499982208,"itl_p95_ms":24.338666697440196,"itl_p99_ms":26.772771790419934,"validation_errors":[],"timestamp":"2026-07-16T14:40:42+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"concurrency":80,"num_prompts":400,"warmup_requests":80,"attempt":1,"gain_pct":16.454156,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c80_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c80_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":480,"failed":0,"duration_s":509.60229949399945,"request_tps":0.9419109773967023,"input_tps":964.5168408542231,"output_tps":3858.0673634168925,"total_tps":4822.584204271116,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":2537.5389005057514,"ttft_p95_ms":4154.587076739699,"ttft_p99_ms":4880.831743057352,"tpot_p50_ms":24.489344787912636,"tpot_p95_ms":25.11027214580131,"tpot_p99_ms":25.309028622957065,"e2e_p50_ms":102768.33627750602,"e2e_p95_ms":104733.21360939881,"e2e_p99_ms":104743.7613113386,"itl_p50_ms":24.15896699676523,"itl_p95_ms":25.35902190429624,"itl_p99_ms":28.550959882268216,"validation_errors":[],"timestamp":"2026-07-16T14:50:01+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"concurrency":96,"num_prompts":480,"warmup_requests":96,"attempt":1,"gain_pct":8.587699,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c96_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c96_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":63.706278053999995,"request_tps":1.2557632064486453,"input_tps":5143.606093613651,"output_tps":160.7376904254266,"total_tps":5304.343784039078,"mean_input_tokens":4096.0,"mean_output_tokens":128.0,"ttft_p50_ms":2722.8040709960624,"ttft_p95_ms":26840.74661414561,"ttft_p99_ms":27275.76444137885,"tpot_p50_ms":20.65726973231667,"tpot_p95_ms":151.34337404136696,"tpot_p99_ms":207.36058258637357,"e2e_p50_ms":5354.688991996227,"e2e_p95_ms":29170.501739544852,"e2e_p99_ms":29177.230598040624,"itl_p50_ms":14.814941496297251,"itl_p95_ms":15.711286505393218,"itl_p99_ms":182.99884561245784,"validation_errors":[],"timestamp":"2026-07-16T14:51:49+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i4096_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i4096_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":13.312623847989016,"request_tps":3.004666882858133,"input_tps":12307.115552186913,"output_tps":384.59736100584104,"total_tps":12691.712913192754,"mean_input_tokens":4096.0,"mean_output_tokens":128.0,"ttft_p50_ms":910.605711498647,"ttft_p95_ms":1522.6708843554661,"ttft_p99_ms":1526.5052215589094,"tpot_p50_ms":12.926063618135059,"tpot_p95_ms":19.92139204688953,"tpot_p99_ms":20.694775890148065,"e2e_p50_ms":2659.4583239930216,"e2e_p95_ms":3033.2144388041343,"e2e_p99_ms":3039.971140041598,"itl_p50_ms":10.341610010073055,"itl_p95_ms":12.366353905235883,"itl_p99_ms":17.325819017132748,"validation_errors":[],"timestamp":"2026-07-16T14:52:41+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i4096_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i4096_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":25.900737181000295,"request_tps":3.088715175979033,"input_tps":12651.377360810118,"output_tps":790.7110850506324,"total_tps":13442.08844586075,"mean_input_tokens":4096.0,"mean_output_tokens":256.0,"ttft_p50_ms":1070.6434219973744,"ttft_p95_ms":2601.758133142721,"ttft_p99_ms":2731.048192037851,"tpot_p50_ms":14.66723954899684,"tpot_p95_ms":20.23464126315529,"tpot_p99_ms":21.929446337492916,"e2e_p50_ms":5262.183128004835,"e2e_p95_ms":6524.885636250838,"e2e_p99_ms":6530.703649207862,"itl_p50_ms":12.11705449532019,"itl_p95_ms":15.253629755170547,"itl_p99_ms":16.63169343562916,"validation_errors":[],"timestamp":"2026-07-16T14:53:51+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i4096_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i4096_o256_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":39.068993989989394,"request_tps":4.0953191689808195,"input_tps":16774.427316145437,"output_tps":1048.4017072590898,"total_tps":17822.829023404527,"mean_input_tokens":4096.0,"mean_output_tokens":256.0,"ttft_p50_ms":2343.7372504995437,"ttft_p95_ms":5162.197779646522,"ttft_p99_ms":5491.549329505941,"tpot_p50_ms":19.213523033335253,"tpot_p95_ms":32.66720932803069,"tpot_p99_ms":34.66762047272469,"e2e_p50_ms":8660.10677999293,"e2e_p95_ms":9275.378481158987,"e2e_p99_ms":9288.702330155647,"itl_p50_ms":14.975201003835537,"itl_p95_ms":17.6850339921657,"itl_p99_ms":28.16523999499614,"validation_errors":[],"timestamp":"2026-07-16T14:55:12+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":256,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":32.589732,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i4096_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i4096_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":37.81364577100612,"request_tps":2.1156383725723837,"input_tps":8665.654774056484,"output_tps":1083.2068467570605,"total_tps":9748.861620813545,"mean_input_tokens":4096.0,"mean_output_tokens":512.0,"ttft_p50_ms":626.4680629974464,"ttft_p95_ms":1079.5922750941827,"ttft_p99_ms":1085.4407174367225,"tpot_p50_ms":12.883891402167945,"tpot_p95_ms":15.779685464180407,"tpot_p99_ms":16.333569430081702,"e2e_p50_ms":7039.921916504682,"e2e_p95_ms":8655.18560165001,"e2e_p99_ms":8659.915602506953,"itl_p50_ms":12.184623003122397,"itl_p95_ms":15.375507500721142,"itl_p99_ms":16.372877460671585,"validation_errors":[],"timestamp":"2026-07-16T14:56:28+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i4096_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i4096_o512_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":56.462142722986755,"request_tps":2.83375713856607,"input_tps":11607.069239566623,"output_tps":1450.8836549458279,"total_tps":13057.952894512451,"mean_input_tokens":4096.0,"mean_output_tokens":512.0,"ttft_p50_ms":1453.6274200072512,"ttft_p95_ms":3663.0988704906445,"ttft_p99_ms":4404.091445246595,"tpot_p50_ms":18.000749555776817,"tpot_p95_ms":23.179044860871425,"tpot_p99_ms":23.890643953811313,"e2e_p50_ms":11132.362890006334,"e2e_p95_ms":13523.678070852475,"e2e_p99_ms":13540.85523293179,"itl_p50_ms":17.186335011501797,"itl_p95_ms":18.29535180295352,"itl_p99_ms":19.591979318647628,"validation_errors":[],"timestamp":"2026-07-16T14:58:04+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":512,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":33.943361,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i4096_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i4096_o512_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":78.62980660999892,"request_tps":3.052277632964196,"input_tps":12502.129184621346,"output_tps":1562.7661480776683,"total_tps":14064.895332699014,"mean_input_tokens":4096.0,"mean_output_tokens":512.0,"ttft_p50_ms":2010.3960574997473,"ttft_p95_ms":8091.9554406515035,"ttft_p99_ms":9461.760722859906,"tpot_p50_ms":23.111368014671182,"tpot_p95_ms":34.88897108324944,"tpot_p99_ms":38.233606207983335,"e2e_p50_ms":14650.032247503987,"e2e_p95_ms":20735.955423545238,"e2e_p99_ms":21034.978971217053,"itl_p50_ms":19.645875494461507,"itl_p95_ms":22.24900509609142,"itl_p99_ms":25.97859771209185,"validation_errors":[],"timestamp":"2026-07-16T15:00:08+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":512,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":7.711335,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i4096_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i4096_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":74.54846192999685,"request_tps":1.0731274385663692,"input_tps":4395.529988367848,"output_tps":1098.882497091962,"total_tps":5494.41248545981,"mean_input_tokens":4096.0,"mean_output_tokens":1024.0,"ttft_p50_ms":664.4156365000526,"ttft_p95_ms":974.0240385996003,"ttft_p99_ms":982.1483760114643,"tpot_p50_ms":14.938762804988556,"tpot_p95_ms":15.543498240374383,"tpot_p99_ms":15.80483324938253,"e2e_p50_ms":16048.860944996704,"e2e_p95_ms":16527.93401440722,"e2e_p99_ms":16534.28529751647,"itl_p50_ms":14.674316000309773,"itl_p95_ms":15.759427801822312,"itl_p99_ms":16.294228279730305,"validation_errors":[],"timestamp":"2026-07-16T15:02:08+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i4096_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i4096_o1024_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":86.79001112899277,"request_tps":1.843530124246648,"input_tps":7551.09938891427,"output_tps":1887.7748472285675,"total_tps":9438.874236142838,"mean_input_tokens":4096.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1054.5200394990388,"ttft_p95_ms":1617.5321672504651,"ttft_p99_ms":1622.538609701296,"tpot_p50_ms":15.406173739000165,"tpot_p95_ms":17.669181664998092,"tpot_p99_ms":18.132834118727168,"e2e_p50_ms":16632.105807497283,"e2e_p95_ms":18881.63431045614,"e2e_p99_ms":18892.737503780663,"itl_p50_ms":15.119137002329808,"itl_p95_ms":17.811838999477914,"itl_p99_ms":18.673117576690856,"validation_errors":[],"timestamp":"2026-07-16T15:04:13+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":1024,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":71.790419,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i4096_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i4096_o1024_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":107.01144343199849,"request_tps":2.242750796577288,"input_tps":9186.307262780572,"output_tps":2296.576815695143,"total_tps":11482.884078475714,"mean_input_tokens":4096.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1429.1901315009454,"ttft_p95_ms":4655.2081622539845,"ttft_p99_ms":5421.287226663553,"tpot_p50_ms":17.961494444764135,"tpot_p95_ms":23.42196408294526,"tpot_p99_ms":23.725152669697756,"e2e_p50_ms":19863.534916497883,"e2e_p95_ms":25332.92310884499,"e2e_p99_ms":25341.139651168924,"itl_p50_ms":17.542485002195463,"itl_p95_ms":19.82468349451665,"itl_p99_ms":22.084278145339344,"validation_errors":[],"timestamp":"2026-07-16T15:06:46+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":1024,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":21.655229,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i4096_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i4096_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":138.75295940799697,"request_tps":0.5765642789986448,"input_tps":2361.607286778449,"output_tps":1180.8036433892246,"total_tps":3542.4109301676735,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":632.4307624963694,"ttft_p95_ms":959.6396220949828,"ttft_p99_ms":969.5783248801308,"tpot_p50_ms":12.087159364439152,"tpot_p95_ms":15.38081124682696,"tpot_p99_ms":15.53437136937437,"e2e_p50_ms":25168.70826199738,"e2e_p95_ms":32143.357299292988,"e2e_p99_ms":32148.356982687113,"itl_p50_ms":12.129640497732908,"itl_p95_ms":15.666113744373433,"itl_p99_ms":16.459487642714517,"validation_errors":[],"timestamp":"2026-07-16T15:09:50+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":2048,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":171.35218937399623,"request_tps":0.9337493765590661,"input_tps":3824.6374463859347,"output_tps":1912.3187231929674,"total_tps":5736.956169578902,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":897.8936719940975,"ttft_p95_ms":1402.46574630728,"ttft_p99_ms":1414.9006537273815,"tpot_p50_ms":17.125081185391775,"tpot_p95_ms":17.535651223030666,"tpot_p99_ms":17.64257730764133,"e2e_p50_ms":35692.707454494666,"e2e_p95_ms":36474.77745984506,"e2e_p99_ms":36482.97110349682,"itl_p50_ms":16.84335399477277,"itl_p95_ms":17.664605999016203,"itl_p99_ms":18.40053380583413,"validation_errors":[],"timestamp":"2026-07-16T15:13:25+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":2048,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":61.950612,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":195.2795769630029,"request_tps":1.2290071687602524,"input_tps":5034.013363241994,"output_tps":2517.006681620997,"total_tps":7551.02004486299,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1402.5059129999136,"ttft_p95_ms":3014.964500940548,"ttft_p99_ms":3645.231739671726,"tpot_p50_ms":17.624665780901058,"tpot_p95_ms":20.07420706065267,"tpot_p99_ms":20.200438402067306,"e2e_p50_ms":38203.180718497606,"e2e_p95_ms":42234.80308325525,"e2e_p99_ms":42259.54538603983,"itl_p50_ms":17.31949599343352,"itl_p95_ms":20.04093239957001,"itl_p99_ms":21.313599089044146,"validation_errors":[],"timestamp":"2026-07-16T15:17:27+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":2048,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":31.620668,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c48_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c48_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":240.90383681698586,"request_tps":1.3283308569431518,"input_tps":5440.84319003915,"output_tps":2720.421595019575,"total_tps":8161.264785058724,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1544.8923770018155,"ttft_p95_ms":9636.169896001238,"ttft_p99_ms":11990.489179552387,"tpot_p50_ms":21.333603897412154,"tpot_p95_ms":26.558363334025465,"tpot_p99_ms":27.72400532863237,"e2e_p50_ms":46171.67401249753,"e2e_p95_ms":58603.167127201596,"e2e_p99_ms":58643.71314777425,"itl_p50_ms":20.934874992235564,"itl_p95_ms":22.243883497139905,"itl_p99_ms":24.41439779868233,"validation_errors":[],"timestamp":"2026-07-16T15:22:09+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":2048,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":8.08162,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c64_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c64_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":274.1387693459983,"request_tps":0.2918230069787383,"input_tps":1195.307036584912,"output_tps":1195.307036584912,"total_tps":2390.614073169824,"mean_input_tokens":4096.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1043.4461850018124,"ttft_p95_ms":1874.445231048594,"ttft_p99_ms":1897.9588461619276,"tpot_p50_ms":11.862804140170747,"tpot_p95_ms":15.333563006644244,"tpot_p99_ms":15.394116480524199,"e2e_p50_ms":49251.109377997636,"e2e_p95_ms":63309.41746450044,"e2e_p99_ms":63314.34459721451,"itl_p50_ms":11.962887001573108,"itl_p95_ms":15.223909802443814,"itl_p99_ms":15.640087766805664,"validation_errors":[],"timestamp":"2026-07-16T15:27:25+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":4096,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i4096_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i4096_o4096_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":361.17721821699524,"request_tps":0.442995825677665,"input_tps":1814.5109019757158,"output_tps":1814.5109019757158,"total_tps":3629.0218039514316,"mean_input_tokens":4096.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1892.501820497273,"ttft_p95_ms":4395.493246797548,"ttft_p99_ms":5224.942043306945,"tpot_p50_ms":17.487011898045463,"tpot_p95_ms":18.118140513893067,"tpot_p99_ms":18.395078147048498,"e2e_p50_ms":74226.02639300021,"e2e_p95_ms":75991.18075425722,"e2e_p99_ms":76000.55798450441,"itl_p50_ms":17.170556006021798,"itl_p95_ms":18.040875489532482,"itl_p99_ms":18.916477599123027,"validation_errors":[],"timestamp":"2026-07-16T15:34:10+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":4096,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":51.802913,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i4096_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i4096_o4096_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":61.62385746999644,"request_tps":1.2981985108438008,"input_tps":21269.684401664832,"output_tps":166.1694093880065,"total_tps":21435.85381105284,"mean_input_tokens":16384.0,"mean_output_tokens":128.0,"ttft_p50_ms":6200.594059002469,"ttft_p95_ms":9982.714110245432,"ttft_p99_ms":12083.678980552644,"tpot_p50_ms":46.56020286221444,"tpot_p95_ms":83.96291618064566,"tpot_p99_ms":94.82308160958443,"e2e_p50_ms":11807.360419501492,"e2e_p95_ms":14518.52319239406,"e2e_p99_ms":15406.82603497858,"itl_p50_ms":14.998138998635113,"itl_p95_ms":19.290680799167607,"itl_p99_ms":913.8215772807598,"validation_errors":[],"timestamp":"2026-07-16T15:35:58+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i16384_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i16384_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":21.187404705997324,"request_tps":1.887914095900456,"input_tps":30931.58454723307,"output_tps":241.65300427525835,"total_tps":31173.237551508326,"mean_input_tokens":16384.0,"mean_output_tokens":128.0,"ttft_p50_ms":1964.1958890060778,"ttft_p95_ms":3285.6321635576023,"ttft_p99_ms":3311.663662179344,"tpot_p50_ms":17.89670796456699,"tpot_p95_ms":32.79861776727342,"tpot_p99_ms":35.39528190300389,"e2e_p50_ms":4186.461981502362,"e2e_p95_ms":4819.31074025124,"e2e_p99_ms":4820.3892862127395,"itl_p50_ms":12.01575299637625,"itl_p95_ms":12.53046725760214,"itl_p99_ms":338.20250410339213,"validation_errors":[],"timestamp":"2026-07-16T15:36:59+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i16384_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i16384_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":48.023830674996134,"request_tps":1.665839623277958,"input_tps":27293.116387786064,"output_tps":426.45494355915724,"total_tps":27719.57133134522,"mean_input_tokens":16384.0,"mean_output_tokens":256.0,"ttft_p50_ms":4218.072452502383,"ttft_p95_ms":7202.70946295495,"ttft_p99_ms":9814.834247549588,"tpot_p50_ms":20.867704264716902,"tpot_p95_ms":43.5013626262719,"tpot_p99_ms":48.00088939475241,"e2e_p50_ms":10424.10009400919,"e2e_p95_ms":13663.989264789414,"e2e_p99_ms":16127.52558444379,"itl_p50_ms":14.975312995375134,"itl_p95_ms":15.900831998442298,"itl_p99_ms":664.3842326034808,"validation_errors":[],"timestamp":"2026-07-16T15:38:27+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i16384_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i16384_o256_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":25.08967021299759,"request_tps":1.5942816171126148,"input_tps":26120.71001477308,"output_tps":408.1360939808294,"total_tps":26528.84610875391,"mean_input_tokens":16384.0,"mean_output_tokens":256.0,"ttft_p50_ms":1583.4694790100912,"ttft_p95_ms":3495.4268617620983,"ttft_p99_ms":3574.4406940348563,"tpot_p50_ms":12.640562578470801,"tpot_p95_ms":19.0468859096262,"tpot_p99_ms":19.132231138046265,"e2e_p50_ms":4888.153373496607,"e2e_p95_ms":6651.8559770600405,"e2e_p99_ms":6667.568836012651,"itl_p50_ms":11.900605997652747,"itl_p95_ms":12.416587599727789,"itl_p99_ms":16.99327580136015,"validation_errors":[],"timestamp":"2026-07-16T15:39:33+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":256,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i16384_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i16384_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":56.001584703000844,"request_tps":1.4285310036184244,"input_tps":23405.051963284266,"output_tps":731.4078738526333,"total_tps":24136.4598371369,"mean_input_tokens":16384.0,"mean_output_tokens":512.0,"ttft_p50_ms":2150.0688629894285,"ttft_p95_ms":4893.693922802049,"ttft_p99_ms":5737.629971327547,"tpot_p50_ms":16.641745551845485,"tpot_p95_ms":22.60100227660779,"tpot_p99_ms":25.976594107668582,"e2e_p50_ms":10941.78693700087,"e2e_p95_ms":14099.559878905711,"e2e_p99_ms":14118.464289340336,"itl_p50_ms":15.120271986234002,"itl_p95_ms":16.094310994958505,"itl_p99_ms":19.090893008979037,"validation_errors":[],"timestamp":"2026-07-16T15:41:11+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i16384_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i16384_o512_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":32.75996512599522,"request_tps":1.221002520795108,"input_tps":20004.90530070705,"output_tps":625.1532906470953,"total_tps":20630.058591354147,"mean_input_tokens":16384.0,"mean_output_tokens":512.0,"ttft_p50_ms":767.6595185039332,"ttft_p95_ms":1176.9666902473525,"ttft_p99_ms":1830.5755027475243,"tpot_p50_ms":11.52257645988111,"tpot_p95_ms":13.118042319375098,"tpot_p99_ms":13.18724245253204,"e2e_p50_ms":6451.215492495976,"e2e_p95_ms":7027.763503049937,"e2e_p99_ms":7034.832992453594,"itl_p50_ms":10.443512990605086,"itl_p95_ms":12.125153494707774,"itl_p99_ms":12.808226600463968,"validation_errors":[],"timestamp":"2026-07-16T15:42:25+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":512,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i16384_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i16384_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":81.79191887499474,"request_tps":0.9780917369387875,"input_tps":16025.055018005094,"output_tps":1001.5659386253184,"total_tps":17026.620956630413,"mean_input_tokens":16384.0,"mean_output_tokens":1024.0,"ttft_p50_ms":999.8386475053849,"ttft_p95_ms":3296.2886180946934,"ttft_p99_ms":4053.4733163932106,"tpot_p50_ms":15.505134818183233,"tpot_p95_ms":18.54402336778967,"tpot_p99_ms":18.632150808406323,"e2e_p50_ms":16877.44059450779,"e2e_p95_ms":19530.405244355643,"e2e_p99_ms":19544.808635528025,"itl_p50_ms":15.109817002667114,"itl_p95_ms":16.077759441395756,"itl_p99_ms":16.771573076403005,"validation_errors":[],"timestamp":"2026-07-16T15:44:38+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i16384_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i16384_o1024_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":165.0628843260056,"request_tps":0.969327542368603,"input_tps":15881.462454167191,"output_tps":992.5914033854494,"total_tps":16874.05385755264,"mean_input_tokens":16384.0,"mean_output_tokens":1024.0,"ttft_p50_ms":4267.802063499403,"ttft_p95_ms":19862.928376002674,"ttft_p99_ms":21657.044113324228,"tpot_p50_ms":25.376833950635163,"tpot_p95_ms":37.8741763598776,"tpot_p99_ms":40.72076095410008,"e2e_p50_ms":35063.07619450672,"e2e_p95_ms":51894.37677649694,"e2e_p99_ms":53688.66714284828,"itl_p50_ms":17.057024502719287,"itl_p95_ms":18.18839350307826,"itl_p99_ms":674.1142390163444,"validation_errors":[],"timestamp":"2026-07-16T15:48:05+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":1024,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":-0.89605,"plateau_streak":1,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i16384_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i16384_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":136.13383251799678,"request_tps":0.5876570028205448,"input_tps":9628.172334211806,"output_tps":1203.5215417764757,"total_tps":10831.693875988281,"mean_input_tokens":16384.0,"mean_output_tokens":2048.0,"ttft_p50_ms":952.042529002938,"ttft_p95_ms":1976.2296090004384,"ttft_p99_ms":2348.7864647468086,"tpot_p50_ms":12.17877617391463,"tpot_p95_ms":15.498852642595702,"tpot_p99_ms":15.918200653907457,"e2e_p50_ms":25627.524990501115,"e2e_p95_ms":32975.65399566083,"e2e_p99_ms":32987.294595711865,"itl_p50_ms":11.984888005827088,"itl_p95_ms":15.32018614525441,"itl_p99_ms":15.952927653561346,"validation_errors":[],"timestamp":"2026-07-16T15:51:05+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":2048,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i16384_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i16384_o2048_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":203.0028582619998,"request_tps":0.7881662424353681,"input_tps":12913.315716061072,"output_tps":1614.164464507634,"total_tps":14527.480180568706,"mean_input_tokens":16384.0,"mean_output_tokens":2048.0,"ttft_p50_ms":2550.8961580053438,"ttft_p95_ms":15365.738819241233,"ttft_p99_ms":18318.440213190624,"tpot_p50_ms":16.89672226160163,"tpot_p95_ms":21.461009534389174,"tpot_p99_ms":22.790276562500463,"e2e_p50_ms":40052.36150848941,"e2e_p95_ms":48792.51678490764,"e2e_p99_ms":48817.223941341945,"itl_p50_ms":14.973600002122112,"itl_p95_ms":17.530049994093133,"itl_p99_ms":18.100957148271835,"validation_errors":[],"timestamp":"2026-07-16T15:55:10+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":2048,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":34.120114,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c32_i16384_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c32_i16384_o2048_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":332.67608038699836,"request_tps":0.24047415704470515,"input_tps":15759.714356081797,"output_tps":30.78069210172226,"total_tps":15790.495048183519,"mean_input_tokens":65536.0,"mean_output_tokens":128.0,"ttft_p50_ms":31602.030366491817,"ttft_p95_ms":50072.25141515082,"ttft_p99_ms":58527.80881931147,"tpot_p50_ms":269.0473995314715,"tpot_p95_ms":453.42705731063467,"tpot_p99_ms":478.3763391067141,"e2e_p50_ms":67748.1715354952,"e2e_p95_ms":83297.61391548891,"e2e_p99_ms":92202.8402509741,"itl_p50_ms":12.720453007204924,"itl_p95_ms":1088.2944416043756,"itl_p99_ms":1121.3325642509153,"validation_errors":[],"timestamp":"2026-07-16T16:01:33+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i65536_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i65536_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":198.8860901500011,"request_tps":0.20112014857264152,"input_tps":13180.610056856634,"output_tps":25.743379017298114,"total_tps":13206.353435873933,"mean_input_tokens":65536.0,"mean_output_tokens":128.0,"ttft_p50_ms":18747.548280000046,"ttft_p95_ms":37954.907318703044,"ttft_p99_ms":40759.29950434613,"tpot_p50_ms":156.92021938971945,"tpot_p95_ms":258.3804646850536,"tpot_p99_ms":286.1677625214699,"e2e_p50_ms":39171.36516800383,"e2e_p95_ms":45190.92419464505,"e2e_p99_ms":50389.63774079078,"itl_p50_ms":12.328808006714098,"itl_p95_ms":904.5139313006074,"itl_p99_ms":1091.3648403182742,"validation_errors":[],"timestamp":"2026-07-16T16:05:39+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i65536_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i65536_o128_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":38.88842260101228,"request_tps":0.12857297019472957,"input_tps":8426.158174681797,"output_tps":16.457340184925386,"total_tps":8442.615514866722,"mean_input_tokens":65536.0,"mean_output_tokens":128.0,"ttft_p50_ms":6736.3390589889605,"ttft_p95_ms":6827.516746404581,"ttft_p99_ms":6834.052311684936,"tpot_p50_ms":7.981791047220794,"tpot_p95_ms":8.017796259758422,"tpot_p99_ms":8.0200000455801,"e2e_p50_ms":7747.252895991551,"e2e_p95_ms":7842.1887119999155,"e2e_p99_ms":7847.936675201636,"itl_p50_ms":8.454770999378525,"itl_p95_ms":8.66287379467394,"itl_p99_ms":10.578372052987115,"validation_errors":[],"timestamp":"2026-07-16T16:07:06+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":128,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c1_i65536_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c1_i65536_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":340.1405706650112,"request_tps":0.23519687711345766,"input_tps":15413.862538507561,"output_tps":60.21040054104516,"total_tps":15474.072939048607,"mean_input_tokens":65536.0,"mean_output_tokens":256.0,"ttft_p50_ms":33961.24007800245,"ttft_p95_ms":49937.02013984801,"ttft_p99_ms":62166.142975594805,"tpot_p50_ms":139.35726555489464,"tpot_p95_ms":236.3478919172656,"tpot_p99_ms":244.71531864295898,"e2e_p50_ms":69747.46793750091,"e2e_p95_ms":78453.58004349907,"e2e_p99_ms":96399.40396136702,"itl_p50_ms":12.44720199611038,"itl_p95_ms":1048.4594318972086,"itl_p99_ms":1099.1333858296275,"validation_errors":[],"timestamp":"2026-07-16T16:13:35+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i65536_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i65536_o256_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":212.1054012759996,"request_tps":0.18858548513788426,"input_tps":12359.138353996383,"output_tps":48.27788419529837,"total_tps":12407.416238191681,"mean_input_tokens":65536.0,"mean_output_tokens":256.0,"ttft_p50_ms":17527.240303992585,"ttft_p95_ms":40766.78363170649,"ttft_p99_ms":45234.88084075011,"tpot_p50_ms":84.98415598624692,"tpot_p95_ms":150.79124716706013,"tpot_p99_ms":159.6644718746755,"e2e_p50_ms":40714.015546502196,"e2e_p95_ms":54715.512818601535,"e2e_p99_ms":68239.62552806508,"itl_p50_ms":12.285736011108384,"itl_p95_ms":823.8078383044922,"itl_p99_ms":1058.4284620790413,"validation_errors":[],"timestamp":"2026-07-16T16:17:54+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":256,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i65536_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i65536_o256_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":44.46872818299744,"request_tps":0.11243856535370274,"input_tps":7368.773819020263,"output_tps":28.7842727305479,"total_tps":7397.5580917508105,"mean_input_tokens":65536.0,"mean_output_tokens":256.0,"ttft_p50_ms":6767.2624370025005,"ttft_p95_ms":6892.302849600674,"ttft_p99_ms":6911.403660320793,"tpot_p50_ms":8.220937709819397,"tpot_p95_ms":8.2882254510004,"tpot_p99_ms":8.295880123954632,"e2e_p50_ms":8871.000218990957,"e2e_p95_ms":9004.320606408874,"e2e_p99_ms":9026.557145289844,"itl_p50_ms":8.483409998007119,"itl_p95_ms":8.675873003085144,"itl_p99_ms":9.017845205380567,"validation_errors":[],"timestamp":"2026-07-16T16:19:29+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":256,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c1_i65536_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c1_i65536_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":365.2319012020016,"request_tps":0.21903891674499099,"input_tps":14354.93444779973,"output_tps":112.14792537343538,"total_tps":14467.082373173163,"mean_input_tokens":65536.0,"mean_output_tokens":512.0,"ttft_p50_ms":33553.05798899644,"ttft_p95_ms":49813.80844564191,"ttft_p99_ms":59237.044557964655,"tpot_p50_ms":79.3469260440206,"tpot_p95_ms":126.30341421105851,"tpot_p99_ms":137.98316977467067,"e2e_p50_ms":73173.0037334928,"e2e_p95_ms":90507.30776954733,"e2e_p99_ms":102572.02681419696,"itl_p50_ms":12.42508299765177,"itl_p95_ms":974.080784598482,"itl_p99_ms":1092.6132275909183,"validation_errors":[],"timestamp":"2026-07-16T16:26:23+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i65536_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i65536_o512_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":221.21875818599074,"request_tps":0.1808164928146365,"input_tps":11849.989673100017,"output_tps":92.57804432109388,"total_tps":11942.56771742111,"mean_input_tokens":65536.0,"mean_output_tokens":512.0,"ttft_p50_ms":19337.339771998813,"ttft_p95_ms":38204.29200109792,"ttft_p99_ms":42388.65437057044,"tpot_p50_ms":48.08991940802787,"tpot_p95_ms":75.00450040077983,"tpot_p99_ms":78.97179943347363,"e2e_p50_ms":44083.91553600086,"e2e_p95_ms":48563.43825120639,"e2e_p99_ms":51796.121535643615,"itl_p50_ms":12.262899006600492,"itl_p95_ms":13.326331894495494,"itl_p99_ms":956.881430233477,"validation_errors":[],"timestamp":"2026-07-16T16:30:51+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":512,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i65536_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i65536_o512_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":55.36560429999372,"request_tps":0.09030877677967596,"input_tps":5918.4759950328435,"output_tps":46.23809371119409,"total_tps":5964.714088744037,"mean_input_tokens":65536.0,"mean_output_tokens":512.0,"ttft_p50_ms":6785.467125999276,"ttft_p95_ms":6867.230469404603,"ttft_p99_ms":6880.844480285305,"tpot_p50_ms":8.349586817987925,"tpot_p95_ms":8.493367793736068,"tpot_p99_ms":8.520835669118448,"e2e_p50_ms":11042.787049998879,"e2e_p95_ms":11206.68308060267,"e2e_p99_ms":11234.859840924619,"itl_p50_ms":8.474283000396099,"itl_p95_ms":8.707773692003684,"itl_p99_ms":9.049925783474462,"validation_errors":[],"timestamp":"2026-07-16T16:32:37+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":512,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c1_i65536_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c1_i65536_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":390.9239437669894,"request_tps":0.2046433872254294,"input_tps":13411.509025205742,"output_tps":209.5548285188397,"total_tps":13621.063853724581,"mean_input_tokens":65536.0,"mean_output_tokens":1024.0,"ttft_p50_ms":34109.15530550119,"ttft_p95_ms":49910.73756714177,"ttft_p99_ms":58476.69062248939,"tpot_p50_ms":42.09774688367832,"tpot_p95_ms":69.21228409907829,"tpot_p99_ms":69.88106847383611,"e2e_p50_ms":79284.3169455009,"e2e_p95_ms":90811.3658934919,"e2e_p99_ms":100775.45445886119,"itl_p50_ms":12.344260001555085,"itl_p95_ms":13.471613003639504,"itl_p99_ms":1075.7671469193883,"validation_errors":[],"timestamp":"2026-07-16T16:39:52+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i65536_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i65536_o1024_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":251.80086339799163,"request_tps":0.15885569040633815,"input_tps":10410.766526469777,"output_tps":162.66822697609027,"total_tps":10573.434753445867,"mean_input_tokens":65536.0,"mean_output_tokens":1024.0,"ttft_p50_ms":17590.520314995956,"ttft_p95_ms":38182.11513094574,"ttft_p99_ms":42436.39091802004,"tpot_p50_ms":30.034462025418435,"tpot_p95_ms":44.99948253655721,"tpot_p99_ms":45.840643626068875,"e2e_p50_ms":50083.116945505026,"e2e_p95_ms":56164.22145070171,"e2e_p99_ms":63398.186792482884,"itl_p50_ms":12.246457001310773,"itl_p95_ms":12.86962560552638,"itl_p99_ms":846.2615448146241,"validation_errors":[],"timestamp":"2026-07-16T16:44:56+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":1024,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i65536_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i65536_o1024_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":76.87229951999325,"request_tps":0.06504293524743045,"input_tps":4262.653804375602,"output_tps":66.60396569336878,"total_tps":4329.257770068971,"mean_input_tokens":65536.0,"mean_output_tokens":1024.0,"ttft_p50_ms":6793.178814987186,"ttft_p95_ms":6830.732189002447,"ttft_p99_ms":6836.927511403337,"tpot_p50_ms":8.389067413488618,"tpot_p95_ms":8.404996331775118,"tpot_p99_ms":8.405754934785953,"e2e_p50_ms":15372.661996996612,"e2e_p95_ms":15412.706287001492,"e2e_p99_ms":15418.935102202231,"itl_p50_ms":8.452879003016278,"itl_p95_ms":8.561783994082361,"itl_p99_ms":8.750601002248002,"validation_errors":[],"timestamp":"2026-07-16T16:46:57+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":1024,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c1_i65536_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c1_i65536_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":782.5810133770137,"request_tps":0.10222583813371851,"input_tps":13398.945055862752,"output_tps":13.084907281115969,"total_tps":13412.029963143868,"mean_input_tokens":131072.0,"mean_output_tokens":128.0,"ttft_p50_ms":58004.37130450155,"ttft_p95_ms":110342.86055744666,"ttft_p99_ms":145009.39971607734,"tpot_p50_ms":726.4293738111247,"tpot_p95_ms":1148.8756013172838,"tpot_p99_ms":1185.2487841417542,"e2e_p50_ms":145932.8185465056,"e2e_p95_ms":251712.73757864692,"e2e_p99_ms":293697.5071116265,"itl_p50_ms":1063.0883960038773,"itl_p95_ms":1341.7487418533708,"itl_p99_ms":1398.4825665256358,"validation_errors":[],"timestamp":"2026-07-16T17:00:59+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":131072,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c16_i131072_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c16_i131072_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":380.95681663800497,"request_tps":0.1049987774283851,"input_tps":13762.399755093293,"output_tps":13.439843510833294,"total_tps":13775.839598604127,"mean_input_tokens":131072.0,"mean_output_tokens":128.0,"ttft_p50_ms":40792.213654494844,"ttft_p95_ms":75418.06702326574,"ttft_p99_ms":76547.42356198141,"tpot_p50_ms":274.08619550392257,"tpot_p95_ms":467.0889975168508,"tpot_p99_ms":530.1239402170978,"e2e_p50_ms":77351.3717395108,"e2e_p95_ms":77859.17398448946,"e2e_p99_ms":82151.91505951836,"itl_p50_ms":10.688919486710802,"itl_p95_ms":1270.379202303593,"itl_p99_ms":1353.2490307680564,"validation_errors":[],"timestamp":"2026-07-16T17:08:21+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":131072,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c8_i131072_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c8_i131072_o128_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":82.18055039300816,"request_tps":0.06084164654639981,"input_tps":7974.636296129715,"output_tps":7.787730757939175,"total_tps":7982.424026887654,"mean_input_tokens":131072.0,"mean_output_tokens":128.0,"ttft_p50_ms":15437.700795999262,"ttft_p95_ms":15571.00856200559,"ttft_p99_ms":15593.639437204693,"tpot_p50_ms":7.764350944867827,"tpot_p95_ms":7.770626307061807,"tpot_p99_ms":7.771325803097836,"e2e_p50_ms":16424.126157013234,"e2e_p95_ms":16554.536877613282,"e2e_p99_ms":16576.65890193428,"itl_p50_ms":8.529996004654095,"itl_p95_ms":8.682068198686466,"itl_p99_ms":9.645803358289447,"validation_errors":[],"timestamp":"2026-07-16T17:10:42+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":131072,"osl":128,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/raw_outputs/vllm_adaptive_c1_i131072_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/logs/vllm_c1_i131072_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":52.02482991799479,"request_tps":1.5377272761122265,"input_tps":1574.63273073892,"output_tps":196.829091342365,"total_tps":1771.461822081285,"mean_input_tokens":1024.0,"mean_output_tokens":128.0,"ttft_p50_ms":1236.4054954814492,"ttft_p95_ms":35677.24587905686,"ttft_p99_ms":35680.42643753841,"tpot_p50_ms":15.189849878005925,"tpot_p95_ms":63.83978625250509,"tpot_p99_ms":164.92346879194884,"e2e_p50_ms":2717.587512001046,"e2e_p95_ms":37134.60250555654,"e2e_p99_ms":37136.07944580115,"itl_p50_ms":11.540331004653126,"itl_p95_ms":11.923325425595976,"itl_p99_ms":14.953987974149642,"validation_errors":[],"timestamp":"2026-07-16T17:21:52+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i1024_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i1024_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":73.70493139498285,"request_tps":0.5427045279459121,"input_tps":555.729436616614,"output_tps":69.46617957707674,"total_tps":625.1956161936906,"mean_input_tokens":1024.0,"mean_output_tokens":128.0,"ttft_p50_ms":3509.9659979896387,"ttft_p95_ms":27374.682503790245,"ttft_p99_ms":32961.56131610186,"tpot_p50_ms":9.04847947639087,"tpot_p95_ms":249.60066149328418,"tpot_p99_ms":278.0204192110943,"e2e_p50_ms":4659.724333003396,"e2e_p95_ms":37692.61776938074,"e2e_p99_ms":37693.20461160998,"itl_p50_ms":9.10900448798202,"itl_p95_ms":9.500342964020094,"itl_p99_ms":10.415470011648722,"validation_errors":[],"timestamp":"2026-07-16T17:24:01+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i1024_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i1024_o128_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":4.607907787984004,"request_tps":1.0850911585163339,"input_tps":1111.1333463207259,"output_tps":138.89166829009073,"total_tps":1250.0250146108165,"mean_input_tokens":1024.0,"mean_output_tokens":128.0,"ttft_p50_ms":68.60028099617921,"ttft_p95_ms":93.32109419628976,"ttft_p99_ms":98.21370284073053,"tpot_p50_ms":6.665402291183118,"tpot_p95_ms":6.668919995271301,"tpot_p99_ms":6.669149680904619,"e2e_p50_ms":914.933813008247,"e2e_p95_ms":940.2739335957449,"e2e_p99_ms":945.1957123156171,"itl_p50_ms":6.641707994276658,"itl_p95_ms":6.847297985223121,"itl_p99_ms":7.190719614736732,"validation_errors":[],"timestamp":"2026-07-16T17:24:40+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":128,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i1024_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i1024_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":61.86757407398545,"request_tps":1.2930844824193455,"input_tps":1324.1185099974098,"output_tps":331.02962749935244,"total_tps":1655.1481374967623,"mean_input_tokens":1024.0,"mean_output_tokens":256.0,"ttft_p50_ms":392.70260799094103,"ttft_p95_ms":27561.146233863838,"ttft_p99_ms":27563.441789576318,"tpot_p50_ms":12.039197992143112,"tpot_p95_ms":70.91977109907734,"tpot_p99_ms":90.06260683440567,"e2e_p50_ms":3515.55855650804,"e2e_p95_ms":30491.65073121112,"e2e_p99_ms":30493.552179136605,"itl_p50_ms":11.53830900148023,"itl_p95_ms":11.843036256323103,"itl_p99_ms":12.743973096075921,"validation_errors":[],"timestamp":"2026-07-16T17:26:16+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i1024_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i1024_o256_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":23.89860760301235,"request_tps":1.6737376781297548,"input_tps":1713.907382404869,"output_tps":428.47684560121724,"total_tps":2142.3842280060862,"mean_input_tokens":1024.0,"mean_output_tokens":256.0,"ttft_p50_ms":240.56512699462473,"ttft_p95_ms":10569.026969805418,"ttft_p99_ms":10732.348576473596,"tpot_p50_ms":9.789249927430944,"tpot_p95_ms":10.445169046887552,"tpot_p99_ms":35.187820401407905,"e2e_p50_ms":2737.936751509551,"e2e_p95_ms":13064.727992002736,"e2e_p99_ms":13066.108254613355,"itl_p50_ms":9.119076014030725,"itl_p95_ms":9.404584451112896,"itl_p99_ms":10.967337382608093,"validation_errors":[],"timestamp":"2026-07-16T17:27:15+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":256,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i1024_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i1024_o256_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":8.878280922974227,"request_tps":0.5631720874095746,"input_tps":576.6882175074044,"output_tps":144.1720543768511,"total_tps":720.8602718842554,"mean_input_tokens":1024.0,"mean_output_tokens":256.0,"ttft_p50_ms":71.51998899644241,"ttft_p95_ms":91.57086079358123,"ttft_p99_ms":95.35931215505116,"tpot_p50_ms":6.661176898028227,"tpot_p95_ms":6.664835591321154,"tpot_p99_ms":6.665386227228879,"e2e_p50_ms":1771.2285799789242,"e2e_p95_ms":1790.1660829898901,"e2e_p99_ms":1793.9073757897131,"itl_p50_ms":6.639207014814019,"itl_p95_ms":6.857857707655057,"itl_p99_ms":7.324532314087264,"validation_errors":[],"timestamp":"2026-07-16T17:27:58+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":256,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i1024_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i1024_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":32.14737898000749,"request_tps":2.488538802797956,"input_tps":2548.2637340651067,"output_tps":1274.1318670325534,"total_tps":3822.39560109766,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":413.3373150107218,"ttft_p95_ms":656.7111001349986,"ttft_p99_ms":658.238969738304,"tpot_p50_ms":11.758502294494425,"tpot_p95_ms":12.275672713020104,"tpot_p99_ms":12.393381951049646,"e2e_p50_ms":6405.572134506656,"e2e_p95_ms":6744.097330357181,"e2e_p99_ms":6745.546151806775,"itl_p50_ms":11.505689995829016,"itl_p95_ms":11.79091099766083,"itl_p99_ms":12.107418500818312,"validation_errors":[],"timestamp":"2026-07-16T17:29:08+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":51.012810627988074,"request_tps":3.1364670566145265,"input_tps":3211.742265973275,"output_tps":1605.8711329866376,"total_tps":4817.613398959913,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":1028.030455505359,"ttft_p95_ms":1825.9766799485078,"ttft_p99_ms":2148.851445445325,"tpot_p50_ms":17.623297716239186,"tpot_p95_ms":19.62757935753655,"tpot_p99_ms":20.25833365699475,"e2e_p50_ms":10080.087392503629,"e2e_p95_ms":11066.638421798416,"e2e_p99_ms":11499.76811571978,"itl_p50_ms":16.898524991120212,"itl_p95_ms":17.623670147440862,"itl_p99_ms":18.77276543091283,"validation_errors":[],"timestamp":"2026-07-16T17:30:43+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":512,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":26.036494,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":85.11057490797248,"request_tps":2.819861107265517,"input_tps":2887.5377738398893,"output_tps":1443.7688869199446,"total_tps":4331.306660759834,"mean_input_tokens":1024.0,"mean_output_tokens":512.0,"ttft_p50_ms":1176.0342860070523,"ttft_p95_ms":32154.562560931667,"ttft_p99_ms":32162.17692289414,"tpot_p50_ms":18.486646824851704,"tpot_p95_ms":74.38846726118646,"tpot_p99_ms":74.39255435188375,"e2e_p50_ms":11688.053546997253,"e2e_p95_ms":41145.71591668209,"e2e_p99_ms":41149.52881608595,"itl_p50_ms":17.28113798890263,"itl_p95_ms":17.941987194353715,"itl_p99_ms":19.419972189934924,"validation_errors":[],"timestamp":"2026-07-16T17:32:54+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":512,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":-10.09435,"plateau_streak":1,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c48_i1024_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c48_i1024_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":61.41648891198565,"request_tps":1.3025817889825302,"input_tps":1333.843751918111,"output_tps":1333.843751918111,"total_tps":2667.687503836222,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":362.0326555101201,"ttft_p95_ms":606.7866882352973,"ttft_p99_ms":607.9597775882576,"tpot_p50_ms":11.630160520515538,"tpot_p95_ms":11.889058939988779,"tpot_p99_ms":11.943324029066188,"e2e_p50_ms":12311.110387017834,"e2e_p95_ms":12538.315513329871,"e2e_p99_ms":12539.58258541039,"itl_p50_ms":11.51326100807637,"itl_p95_ms":11.765323003055528,"itl_p99_ms":11.965371266705917,"validation_errors":[],"timestamp":"2026-07-16T17:34:33+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":80.85768987398478,"request_tps":1.9787851996434358,"input_tps":2026.2760444348783,"output_tps":2026.2760444348783,"total_tps":4052.5520888697565,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":601.8352310056798,"ttft_p95_ms":875.1060022099409,"ttft_p99_ms":879.6998479493777,"tpot_p50_ms":15.16937012902533,"tpot_p95_ms":15.621438565148944,"tpot_p99_ms":15.709627801430935,"e2e_p50_ms":16193.191637503332,"e2e_p95_ms":16394.8902862423,"e2e_p99_ms":16400.64757927088,"itl_p50_ms":15.000293991761282,"itl_p95_ms":15.395940997404976,"itl_p99_ms":15.808226090157401,"validation_errors":[],"timestamp":"2026-07-16T17:36:30+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":1024,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":51.912549,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":94.47741836198838,"request_tps":2.5402895650730493,"input_tps":2601.2565146348024,"output_tps":2601.2565146348024,"total_tps":5202.513029269605,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":872.004470002139,"ttft_p95_ms":1068.056017033814,"ttft_p99_ms":1074.3570806449861,"tpot_p50_ms":17.694852847022954,"tpot_p95_ms":18.194023427918026,"tpot_p99_ms":18.255092446847165,"e2e_p50_ms":18907.909675486735,"e2e_p95_ms":19168.525980616687,"e2e_p99_ms":19178.07574301085,"itl_p50_ms":17.358476005028933,"itl_p95_ms":17.918874911265448,"itl_p99_ms":18.629112805938348,"validation_errors":[],"timestamp":"2026-07-16T17:38:40+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":1024,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":28.376216,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c48_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c48_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":111.18827333900845,"request_tps":2.8780013430403155,"input_tps":2947.073375273283,"output_tps":2947.073375273283,"total_tps":5894.146750546566,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1236.4594759856118,"ttft_p95_ms":3486.339133232832,"ttft_p99_ms":3954.5976989151677,"tpot_p50_ms":20.184367791299923,"tpot_p95_ms":22.733724067629,"tpot_p99_ms":23.26845901167211,"e2e_p50_ms":21600.943245503004,"e2e_p95_ms":24639.851032342995,"e2e_p99_ms":24654.7415991014,"itl_p50_ms":19.554279977455735,"itl_p95_ms":20.101924979826435,"itl_p99_ms":21.4101209945511,"validation_errors":[],"timestamp":"2026-07-16T17:41:09+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":1024,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":13.294224,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c64_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c64_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":400,"failed":0,"duration_s":131.46577305399114,"request_tps":3.0426170303332536,"input_tps":3115.6398390612517,"output_tps":3115.6398390612517,"total_tps":6231.279678122503,"mean_input_tokens":1024.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1146.1891874932917,"ttft_p95_ms":4487.726777615899,"ttft_p99_ms":5507.241578921676,"tpot_p50_ms":23.907261485326536,"tpot_p95_ms":26.730242256350078,"tpot_p99_ms":27.79943066497788,"e2e_p50_ms":25425.060303488863,"e2e_p95_ms":29093.009902854097,"e2e_p99_ms":29212.41997377714,"itl_p50_ms":23.131843001465313,"itl_p95_ms":27.57070749357808,"itl_p99_ms":28.485130463086534,"validation_errors":[],"timestamp":"2026-07-16T17:44:07+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":1024,"concurrency":80,"num_prompts":400,"warmup_requests":80,"attempt":1,"gain_pct":5.719792,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c80_i1024_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c80_i1024_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":123.58838067599572,"request_tps":0.6473100429216823,"input_tps":662.8454839518026,"output_tps":1325.6909679036053,"total_tps":1988.536451855408,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":503.70064900198486,"ttft_p95_ms":3332.6880413966137,"ttft_p99_ms":3334.5142474136082,"tpot_p50_ms":11.566837329992605,"tpot_p95_ms":11.80489444616316,"tpot_p99_ms":12.187286488214971,"e2e_p50_ms":24182.123156497255,"e2e_p95_ms":27096.394883874746,"e2e_p99_ms":27099.026713256026,"itl_p50_ms":11.55229500727728,"itl_p95_ms":11.839765001786873,"itl_p99_ms":12.08073129178956,"validation_errors":[],"timestamp":"2026-07-16T17:46:50+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":2048,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":158.5647476890008,"request_tps":1.0090515220559253,"input_tps":1033.2687585852675,"output_tps":2066.537517170535,"total_tps":3099.8062757558023,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":669.473130008555,"ttft_p95_ms":865.0462039600825,"ttft_p99_ms":869.1866847060737,"tpot_p50_ms":15.184576202977428,"tpot_p95_ms":15.339674701513589,"tpot_p99_ms":15.58220753159747,"e2e_p50_ms":31597.606307492242,"e2e_p95_ms":32205.554889343333,"e2e_p99_ms":32214.89342865796,"itl_p50_ms":15.059143508551642,"itl_p95_ms":15.587451988540124,"itl_p99_ms":16.146498053567488,"validation_errors":[],"timestamp":"2026-07-16T17:50:08+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":2048,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":55.883805,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":187.52933345900965,"request_tps":1.279799781576355,"input_tps":1310.5149763341876,"output_tps":2621.0299526683752,"total_tps":3931.5449290025626,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":996.749361496768,"ttft_p95_ms":2177.432288348791,"ttft_p99_ms":2548.245648367156,"tpot_p50_ms":17.70298705446327,"tpot_p95_ms":18.496204302150304,"tpot_p99_ms":18.711075803456517,"e2e_p50_ms":37275.27416749217,"e2e_p95_ms":39046.80715519644,"e2e_p99_ms":39530.66943249112,"itl_p50_ms":17.395492002833635,"itl_p95_ms":17.924149986356497,"itl_p99_ms":18.54355659452267,"validation_errors":[],"timestamp":"2026-07-16T17:53:52+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":2048,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":26.831956,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c48_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c48_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":231.3560413280211,"request_tps":1.3831495307541928,"input_tps":1416.3451194922934,"output_tps":2832.690238984587,"total_tps":4249.03535847688,"mean_input_tokens":1024.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1237.1772454935126,"ttft_p95_ms":4075.03496848949,"ttft_p99_ms":4438.276356909774,"tpot_p50_ms":19.961317755979394,"tpot_p95_ms":29.797527954492573,"tpot_p99_ms":29.79867891234008,"e2e_p50_ms":41953.71764700394,"e2e_p95_ms":61770.79950050393,"e2e_p99_ms":61851.04672904883,"itl_p50_ms":19.631187984487042,"itl_p95_ms":20.16204949177336,"itl_p99_ms":20.984756899997592,"validation_errors":[],"timestamp":"2026-07-16T17:58:42+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":2048,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":8.075462,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c64_i1024_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c64_i1024_o2048_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":238.9891159539984,"request_tps":0.3347432776620619,"input_tps":342.7771163259514,"output_tps":1371.1084653038056,"total_tps":1713.885581629757,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":407.33929100679234,"ttft_p95_ms":588.6433666979428,"ttft_p99_ms":591.8217892898247,"tpot_p50_ms":11.581026416729486,"tpot_p95_ms":11.690784101489584,"tpot_p99_ms":11.69744405043216,"e2e_p50_ms":47807.511286999215,"e2e_p95_ms":48180.03251795162,"e2e_p99_ms":48181.1768043015,"itl_p50_ms":11.548227979801595,"itl_p95_ms":11.81682851165533,"itl_p99_ms":11.96142429544125,"validation_errors":[],"timestamp":"2026-07-16T18:03:17+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":4096,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":315.3864424380008,"request_tps":0.507314134251199,"input_tps":519.4896734732278,"output_tps":2077.958693892911,"total_tps":2597.448367366139,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":827.0409834949533,"ttft_p95_ms":2127.0444475507247,"ttft_p99_ms":2149.356553926773,"tpot_p50_ms":15.163013353477309,"tpot_p95_ms":15.352130105484576,"tpot_p99_ms":15.488373952998492,"e2e_p50_ms":63035.64160001406,"e2e_p95_ms":63718.42900119955,"e2e_p99_ms":63728.90193682426,"itl_p50_ms":15.009783499408513,"itl_p95_ms":15.42943195818225,"itl_p99_ms":15.688629885262342,"validation_errors":[],"timestamp":"2026-07-16T18:09:14+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":4096,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":51.553196,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":384.594265947002,"request_tps":0.6240342648090146,"input_tps":639.0110871644309,"output_tps":2556.0443486577237,"total_tps":3195.0554358221543,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":988.36895050772,"ttft_p95_ms":2441.5721719269623,"ttft_p99_ms":5020.39980761183,"tpot_p50_ms":17.599965752136836,"tpot_p95_ms":22.463060006479317,"tpot_p99_ms":22.464100362125627,"e2e_p50_ms":73127.70336250833,"e2e_p95_ms":92961.33970778319,"e2e_p99_ms":93251.61461467971,"itl_p50_ms":17.415053996955976,"itl_p95_ms":17.90137597708963,"itl_p99_ms":18.575296591734517,"validation_errors":[],"timestamp":"2026-07-16T18:16:19+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":4096,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":23.007467,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c48_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c48_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":320,"failed":0,"duration_s":413.36956618897966,"request_tps":0.7741256884250303,"input_tps":792.704704947231,"output_tps":3170.818819788924,"total_tps":3963.523524736155,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1353.9671769976849,"ttft_p95_ms":3372.6759848868824,"ttft_p99_ms":3923.934282742266,"tpot_p50_ms":19.719561329914566,"tpot_p95_ms":20.46792597871924,"tpot_p99_ms":20.599813236414754,"e2e_p50_ms":82000.917450001,"e2e_p95_ms":85134.73726001248,"e2e_p99_ms":85152.08560849074,"itl_p50_ms":19.557946507120505,"itl_p95_ms":20.046211070439313,"itl_p99_ms":20.78921643726062,"validation_errors":[],"timestamp":"2026-07-16T18:24:00+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":4096,"concurrency":64,"num_prompts":320,"warmup_requests":64,"attempt":1,"gain_pct":24.051792,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c64_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c64_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":400,"failed":0,"duration_s":501.7031515800045,"request_tps":0.797284208281904,"input_tps":816.4190292806697,"output_tps":3265.676117122679,"total_tps":4082.095146403349,"mean_input_tokens":1024.0,"mean_output_tokens":4096.0,"ttft_p50_ms":2250.2203854965046,"ttft_p95_ms":4658.4656137551065,"ttft_p99_ms":5716.611436419189,"tpot_p50_ms":24.059723582905082,"tpot_p95_ms":24.354225822548287,"tpot_p99_ms":24.487317885165243,"e2e_p50_ms":99781.74352001224,"e2e_p95_ms":103402.10737639281,"e2e_p99_ms":104522.76062945835,"itl_p50_ms":23.14190100878477,"itl_p95_ms":23.698028920625802,"itl_p99_ms":24.58483098977012,"validation_errors":[],"timestamp":"2026-07-16T18:33:07+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":4096,"concurrency":80,"num_prompts":400,"warmup_requests":80,"attempt":1,"gain_pct":2.991571,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c80_i1024_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c80_i1024_o4096_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":31.652223498997046,"request_tps":2.527468567967585,"input_tps":10352.511254395227,"output_tps":323.51597669985085,"total_tps":10676.027231095079,"mean_input_tokens":4096.0,"mean_output_tokens":128.0,"ttft_p50_ms":2794.544570497237,"ttft_p95_ms":4019.1005029933863,"ttft_p99_ms":4515.16735956771,"tpot_p50_ms":28.070199511753664,"tpot_p95_ms":46.8207419802839,"tpot_p99_ms":47.12049638943226,"e2e_p50_ms":6348.330602006172,"e2e_p95_ms":6970.749582740245,"e2e_p99_ms":7275.766055090349,"itl_p50_ms":11.53999000962358,"itl_p95_ms":12.58592710510129,"itl_p99_ms":611.4497984829359,"validation_errors":[],"timestamp":"2026-07-16T18:34:16+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i4096_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i4096_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":8.121882901003119,"request_tps":4.924966351713797,"input_tps":20172.662176619713,"output_tps":630.395693019366,"total_tps":20803.05786963908,"mean_input_tokens":4096.0,"mean_output_tokens":128.0,"ttft_p50_ms":391.37257999391295,"ttft_p95_ms":654.7860209611827,"ttft_p99_ms":655.5212389715598,"tpot_p50_ms":9.02552549213017,"tpot_p95_ms":11.834378843241568,"tpot_p99_ms":12.646475187780089,"e2e_p50_ms":1594.1091245040298,"e2e_p95_ms":1801.0780599128339,"e2e_p99_ms":1801.8622757561388,"itl_p50_ms":9.091041501960717,"itl_p95_ms":9.295591751288157,"itl_p99_ms":11.05267516104506,"validation_errors":[],"timestamp":"2026-07-16T18:35:04+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i4096_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i4096_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":17.724754639988532,"request_tps":4.513461631762919,"input_tps":18487.138843700915,"output_tps":1155.4461777313072,"total_tps":19642.58502143222,"mean_input_tokens":4096.0,"mean_output_tokens":256.0,"ttft_p50_ms":584.0600009978516,"ttft_p95_ms":737.9011562516098,"ttft_p99_ms":741.6909766758907,"tpot_p50_ms":11.493453029349135,"tpot_p95_ms":13.185771285709194,"tpot_p99_ms":13.456647717705515,"e2e_p50_ms":3522.1422600006917,"e2e_p95_ms":3669.390212147846,"e2e_p99_ms":3672.551417663053,"itl_p50_ms":11.582269973587245,"itl_p95_ms":11.904317999142222,"itl_p99_ms":12.9585687990766,"validation_errors":[],"timestamp":"2026-07-16T18:35:55+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i4096_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i4096_o256_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":46.13867022198974,"request_tps":3.467806922700252,"input_tps":14204.137155380233,"output_tps":887.7585722112645,"total_tps":15091.895727591496,"mean_input_tokens":4096.0,"mean_output_tokens":256.0,"ttft_p50_ms":1084.7839874913916,"ttft_p95_ms":6128.039478484425,"ttft_p99_ms":6135.151792074903,"tpot_p50_ms":28.80620697841925,"tpot_p95_ms":44.439678273264136,"tpot_p99_ms":49.92359239300004,"e2e_p50_ms":9273.995624491363,"e2e_p95_ms":15928.954499101383,"e2e_p99_ms":15931.015178921225,"itl_p50_ms":14.997396006947383,"itl_p95_ms":15.599585502059199,"itl_p99_ms":613.4823884931393,"validation_errors":[],"timestamp":"2026-07-16T18:37:16+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":256,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":-23.167466,"plateau_streak":1,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i4096_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i4096_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":32.73595824901713,"request_tps":2.4437958831524944,"input_tps":10009.787937392617,"output_tps":1251.2234921740771,"total_tps":11261.011429566694,"mean_input_tokens":4096.0,"mean_output_tokens":512.0,"ttft_p50_ms":580.2664404909592,"ttft_p95_ms":736.464851902565,"ttft_p99_ms":739.5049455872504,"tpot_p50_ms":11.585709298447135,"tpot_p95_ms":12.479319499847486,"tpot_p99_ms":12.621792366471547,"e2e_p50_ms":6533.585163517273,"e2e_p95_ms":6650.0653961644275,"e2e_p99_ms":6653.992265443085,"itl_p50_ms":11.581516009755433,"itl_p95_ms":11.963144004403148,"itl_p99_ms":12.631978923745926,"validation_errors":[],"timestamp":"2026-07-16T18:38:24+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i4096_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i4096_o512_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":43.90981093500159,"request_tps":3.6438325875928577,"input_tps":14925.138278780345,"output_tps":1865.6422848475431,"total_tps":16790.78056362789,"mean_input_tokens":4096.0,"mean_output_tokens":512.0,"ttft_p50_ms":799.3582674971549,"ttft_p95_ms":1175.5077986585093,"ttft_p99_ms":1183.6164445849136,"tpot_p50_ms":15.426591234855762,"tpot_p95_ms":17.008792841357067,"tpot_p99_ms":17.19561321969741,"e2e_p50_ms":8660.748706999584,"e2e_p95_ms":9361.31468063686,"e2e_p99_ms":9369.353665226081,"itl_p50_ms":15.0350559852086,"itl_p95_ms":17.14442181983031,"itl_p99_ms":17.799102468416095,"validation_errors":[],"timestamp":"2026-07-16T18:39:52+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":512,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":49.105439,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i4096_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i4096_o512_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":76.34642335801618,"request_tps":3.1435657290001995,"input_tps":12876.045225984817,"output_tps":1609.5056532481021,"total_tps":14485.55087923292,"mean_input_tokens":4096.0,"mean_output_tokens":512.0,"ttft_p50_ms":1696.7760205006925,"ttft_p95_ms":7079.009777032479,"ttft_p99_ms":9906.125367596096,"tpot_p50_ms":18.908712023488928,"tpot_p95_ms":37.75210648097284,"tpot_p99_ms":42.41224298877765,"e2e_p50_ms":13078.812665509759,"e2e_p95_ms":24145.498285339272,"e2e_p99_ms":26969.445044811862,"itl_p50_ms":17.41874399886001,"itl_p95_ms":17.961374601873104,"itl_p99_ms":383.83775273512606,"validation_errors":[],"timestamp":"2026-07-16T18:41:51+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":512,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":-13.729139,"plateau_streak":1,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c48_i4096_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c48_i4096_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":62.43282029798138,"request_tps":1.2813773207452974,"input_tps":5248.521505772738,"output_tps":1312.1303764431846,"total_tps":6560.651882215922,"mean_input_tokens":4096.0,"mean_output_tokens":1024.0,"ttft_p50_ms":563.8070285058348,"ttft_p95_ms":777.6129835590837,"ttft_p99_ms":779.2977576740668,"tpot_p50_ms":11.56457677565038,"tpot_p95_ms":12.043608254741104,"tpot_p99_ms":12.209488064032353,"e2e_p50_ms":12494.566029999987,"e2e_p95_ms":12608.202403750329,"e2e_p99_ms":12608.88043828163,"itl_p50_ms":11.578381003346294,"itl_p95_ms":11.835356391384266,"itl_p99_ms":12.015851708129048,"validation_errors":[],"timestamp":"2026-07-16T18:43:33+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i4096_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i4096_o1024_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":103.49840845502331,"request_tps":1.5459174917605643,"input_tps":6332.078046251271,"output_tps":1583.0195115628178,"total_tps":7915.097557814089,"mean_input_tokens":4096.0,"mean_output_tokens":1024.0,"ttft_p50_ms":912.6817460055463,"ttft_p95_ms":22953.754147146537,"ttft_p99_ms":22989.648280964466,"tpot_p50_ms":15.027860665688419,"tpot_p95_ms":15.874237878655665,"tpot_p99_ms":34.52500240562136,"e2e_p50_ms":16423.773564994917,"e2e_p95_ms":38356.92362101836,"e2e_p99_ms":38391.64368835016,"itl_p50_ms":15.018527017673478,"itl_p95_ms":15.399335196707398,"itl_p99_ms":15.800710814073682,"validation_errors":[],"timestamp":"2026-07-16T18:45:53+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":1024,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":20.644986,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i4096_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i4096_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":121.8976368480071,"request_tps":0.6562883585655658,"input_tps":2688.1571166845574,"output_tps":1344.0785583422787,"total_tps":4032.235675026836,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":452.4103995063342,"ttft_p95_ms":869.4939421853633,"ttft_p99_ms":873.1171497041942,"tpot_p50_ms":11.663301339766598,"tpot_p95_ms":11.80714733549408,"tpot_p99_ms":11.878064581100528,"e2e_p50_ms":24396.55241399305,"e2e_p95_ms":24617.466119487653,"e2e_p99_ms":24621.38425450714,"itl_p50_ms":11.577293000300415,"itl_p95_ms":11.859655751322862,"itl_p99_ms":12.521442596334962,"validation_errors":[],"timestamp":"2026-07-16T18:48:39+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":2048,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":158.23376187801477,"request_tps":1.0111622077426614,"input_tps":4141.720402913941,"output_tps":2070.8602014569706,"total_tps":6212.580604370912,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":756.7443619918777,"ttft_p95_ms":1068.711685812741,"ttft_p99_ms":1076.5162225518725,"tpot_p50_ms":15.088843866877845,"tpot_p95_ms":15.323755214356261,"tpot_p99_ms":15.423469172625605,"e2e_p50_ms":31670.834326505428,"e2e_p95_ms":31954.730625603406,"e2e_p99_ms":31962.703760440054,"itl_p50_ms":14.991424977779388,"itl_p95_ms":15.389906987547874,"itl_p99_ms":15.759109035134308,"validation_errors":[],"timestamp":"2026-07-16T18:52:01+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":2048,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":54.072854,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":240,"failed":0,"duration_s":206.79596914799185,"request_tps":1.1605642072657902,"input_tps":4753.670992960677,"output_tps":2376.8354964803384,"total_tps":7130.506489441016,"mean_input_tokens":4096.0,"mean_output_tokens":2048.0,"ttft_p50_ms":1145.1450809981907,"ttft_p95_ms":6598.897911023221,"ttft_p99_ms":9627.660050078412,"tpot_p50_ms":17.8303520544688,"tpot_p95_ms":22.353743953683303,"tpot_p99_ms":23.511489080978755,"e2e_p50_ms":37381.53092398716,"e2e_p95_ms":50990.53589172253,"e2e_p99_ms":54006.87992766645,"itl_p50_ms":17.39579500281252,"itl_p95_ms":17.878196993842717,"itl_p99_ms":18.62087064306253,"validation_errors":[],"timestamp":"2026-07-16T18:56:11+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":2048,"concurrency":48,"num_prompts":240,"warmup_requests":48,"attempt":1,"gain_pct":14.775275,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c48_i4096_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c48_i4096_o2048_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":239.23693925398402,"request_tps":0.3343965202425058,"input_tps":1369.6881469133039,"output_tps":1369.6881469133039,"total_tps":2739.3762938266077,"mean_input_tokens":4096.0,"mean_output_tokens":4096.0,"ttft_p50_ms":535.376151005039,"ttft_p95_ms":764.6692137466744,"ttft_p99_ms":770.1709890385973,"tpot_p50_ms":11.550237434308961,"tpot_p95_ms":11.648775638909902,"tpot_p99_ms":11.68880080481187,"e2e_p50_ms":47866.3036479993,"e2e_p95_ms":48047.369737493864,"e2e_p99_ms":48052.40901362762,"itl_p50_ms":11.535027006175369,"itl_p95_ms":11.756750987842679,"itl_p99_ms":11.894877816084774,"validation_errors":[],"timestamp":"2026-07-16T19:00:46+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":4096,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i4096_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i4096_o4096_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":333.05284896001103,"request_tps":0.4804042376446114,"input_tps":1967.7357573923282,"output_tps":1967.7357573923282,"total_tps":3935.4715147846564,"mean_input_tokens":4096.0,"mean_output_tokens":4096.0,"ttft_p50_ms":1095.2641390031204,"ttft_p95_ms":6705.06853362167,"ttft_p99_ms":7314.000860454107,"tpot_p50_ms":15.522370325397588,"tpot_p95_ms":16.71266772969576,"tpot_p99_ms":17.02914903613981,"e2e_p50_ms":66306.44073350413,"e2e_p95_ms":73349.6555153004,"e2e_p99_ms":73959.53097622347,"itl_p50_ms":14.9437389918603,"itl_p95_ms":15.368272416526452,"itl_p99_ms":15.751762358704575,"validation_errors":[],"timestamp":"2026-07-16T19:07:01+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":4096,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":43.663049,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i4096_o4096_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i4096_o4096_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":86.4904059410037,"request_tps":0.924958082108773,"input_tps":15154.513217270136,"output_tps":118.39463450992294,"total_tps":15272.907851780057,"mean_input_tokens":16384.0,"mean_output_tokens":128.0,"ttft_p50_ms":7593.278043495957,"ttft_p95_ms":12824.893658250221,"ttft_p99_ms":14977.241600849084,"tpot_p50_ms":78.05374694479846,"tpot_p95_ms":140.04996209068491,"tpot_p99_ms":159.51224197268965,"e2e_p50_ms":17123.760913003935,"e2e_p95_ms":26093.928836856503,"e2e_p99_ms":26102.272943213466,"itl_p50_ms":11.707582991220988,"itl_p95_ms":648.0138543454814,"itl_p99_ms":666.6313856895431,"validation_errors":[],"timestamp":"2026-07-16T19:09:09+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i16384_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i16384_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":8.840474782977253,"request_tps":4.524643866076273,"input_tps":74131.76510179366,"output_tps":579.154414857763,"total_tps":74710.91951665143,"mean_input_tokens":16384.0,"mean_output_tokens":128.0,"ttft_p50_ms":611.7119685077341,"ttft_p95_ms":705.7421804507612,"ttft_p99_ms":728.9413817677995,"tpot_p50_ms":8.631740236223134,"tpot_p95_ms":11.94178506870173,"tpot_p99_ms":12.782024573981596,"e2e_p50_ms":1712.7858160092728,"e2e_p95_ms":1891.0789757341263,"e2e_p99_ms":1892.5092905788915,"itl_p50_ms":9.147118500550278,"itl_p95_ms":9.401795730809681,"itl_p99_ms":14.493186495383268,"validation_errors":[],"timestamp":"2026-07-16T19:09:55+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i16384_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i16384_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":20.272567973006517,"request_tps":3.946219349542801,"input_tps":64654.85782290925,"output_tps":1010.2321534829571,"total_tps":65665.0899763922,"mean_input_tokens":16384.0,"mean_output_tokens":256.0,"ttft_p50_ms":732.9621660028351,"ttft_p95_ms":1059.6370970582939,"ttft_p99_ms":1330.065922466165,"tpot_p50_ms":12.931638015616757,"tpot_p95_ms":14.641560330985836,"tpot_p99_ms":15.307604808072721,"e2e_p50_ms":3975.734623003518,"e2e_p95_ms":4291.209159711434,"e2e_p99_ms":4300.588087121723,"itl_p50_ms":11.725543983629905,"itl_p95_ms":12.27384575031465,"itl_p99_ms":21.68184940965146,"validation_errors":[],"timestamp":"2026-07-16T19:10:50+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i16384_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i16384_o256_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":125.27427309198538,"request_tps":1.2771975925377472,"input_tps":20925.60535613845,"output_tps":326.9625836896633,"total_tps":21252.567939828114,"mean_input_tokens":16384.0,"mean_output_tokens":256.0,"ttft_p50_ms":1955.9702225087676,"ttft_p95_ms":25687.426767007855,"ttft_p99_ms":26337.214467667218,"tpot_p50_ms":71.67799913920189,"tpot_p95_ms":145.99830021673134,"tpot_p99_ms":162.3011558566975,"e2e_p50_ms":24834.69372948457,"e2e_p95_ms":61469.757113161904,"e2e_p99_ms":62129.46268044296,"itl_p50_ms":15.310183982364833,"itl_p95_ms":637.7874485973734,"itl_p99_ms":661.2853236787487,"validation_errors":[],"timestamp":"2026-07-16T19:13:39+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":256,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":-67.634906,"plateau_streak":1,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i16384_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i16384_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":130.2362937349826,"request_tps":0.6142680945972843,"input_tps":10064.168461881905,"output_tps":314.50526443380954,"total_tps":10378.673726315716,"mean_input_tokens":16384.0,"mean_output_tokens":512.0,"ttft_p50_ms":8992.384835495614,"ttft_p95_ms":14412.819043596388,"ttft_p99_ms":18462.384867458015,"tpot_p50_ms":34.2774762993976,"tpot_p95_ms":47.27433437131593,"tpot_p99_ms":48.483385741755086,"e2e_p50_ms":26175.92193999735,"e2e_p95_ms":32332.358848227883,"e2e_p99_ms":36386.61295437544,"itl_p50_ms":11.672908993205056,"itl_p95_ms":13.615285683772521,"itl_p99_ms":648.6071743268986,"validation_errors":[],"timestamp":"2026-07-16T19:16:28+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i16384_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i16384_o512_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":73.81995394398109,"request_tps":0.5418589129756752,"input_tps":8877.816430193463,"output_tps":277.4317634435457,"total_tps":9155.248193637008,"mean_input_tokens":16384.0,"mean_output_tokens":512.0,"ttft_p50_ms":5848.043638005038,"ttft_p95_ms":10270.304282526195,"ttft_p99_ms":10305.013074664457,"tpot_p50_ms":17.2701541448243,"tpot_p95_ms":26.505919335452827,"tpot_p99_ms":26.557917271059214,"e2e_p50_ms":14963.280752504943,"e2e_p95_ms":15029.205415995966,"e2e_p99_ms":15033.096053168701,"itl_p50_ms":9.254616481484845,"itl_p95_ms":9.525524696800858,"itl_p99_ms":610.016989769938,"validation_errors":[],"timestamp":"2026-07-16T19:18:22+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":512,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i16384_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i16384_o512_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":18.15804166198359,"request_tps":0.2753600907562737,"input_tps":4511.4997269507885,"output_tps":140.98436646721214,"total_tps":4652.484093418,"mean_input_tokens":16384.0,"mean_output_tokens":512.0,"ttft_p50_ms":186.52946400106885,"ttft_p95_ms":232.34625218901783,"ttft_p99_ms":237.30703042820096,"tpot_p50_ms":6.7337785733917475,"tpot_p95_ms":6.740160819146246,"tpot_p99_ms":6.740686006462676,"e2e_p50_ms":3627.490315004252,"e2e_p95_ms":3669.8924877971876,"e2e_p99_ms":3676.717489558505,"itl_p50_ms":6.754819012712687,"itl_p95_ms":6.876269215717912,"itl_p99_ms":7.019366837339476,"validation_errors":[],"timestamp":"2026-07-16T19:19:23+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":512,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i16384_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i16384_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":113.21908437900129,"request_tps":0.7065946561818122,"input_tps":11576.84684688281,"output_tps":723.5529279301757,"total_tps":12300.399774812988,"mean_input_tokens":16384.0,"mean_output_tokens":1024.0,"ttft_p50_ms":1313.6420960072428,"ttft_p95_ms":15357.72944069031,"ttft_p99_ms":15506.10770056053,"tpot_p50_ms":15.984088203331268,"tpot_p95_ms":27.630180807318535,"tpot_p99_ms":30.014112934037232,"e2e_p50_ms":22529.161178987124,"e2e_p95_ms":37872.18460994045,"e2e_p99_ms":38033.13407269189,"itl_p50_ms":11.689694976666942,"itl_p95_ms":12.091477410285734,"itl_p99_ms":100.4964998154901,"validation_errors":[],"timestamp":"2026-07-16T19:21:55+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i16384_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i16384_o1024_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":50.61696275198483,"request_tps":0.7902489170674606,"input_tps":12947.438257233274,"output_tps":809.2148910770796,"total_tps":13756.653148310354,"mean_input_tokens":16384.0,"mean_output_tokens":1024.0,"ttft_p50_ms":573.4946330048842,"ttft_p95_ms":733.3654505957384,"ttft_p99_ms":790.053364574851,"tpot_p50_ms":9.346392439878139,"tpot_p95_ms":9.664219554327293,"tpot_p99_ms":9.85550387294379,"e2e_p50_ms":10074.975329000154,"e2e_p95_ms":10331.053190038074,"e2e_p99_ms":10332.716352115385,"itl_p50_ms":9.268547495594248,"itl_p95_ms":9.488147572847083,"itl_p99_ms":10.121553320495892,"validation_errors":[],"timestamp":"2026-07-16T19:23:23+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":1024,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i16384_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i16384_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":125.30134890798945,"request_tps":0.6384608042707116,"input_tps":10460.541817171339,"output_tps":1307.5677271464174,"total_tps":11768.109544317756,"mean_input_tokens":16384.0,"mean_output_tokens":2048.0,"ttft_p50_ms":779.641143992194,"ttft_p95_ms":1110.0852024552294,"ttft_p99_ms":1200.1118386295275,"tpot_p50_ms":11.868535543230175,"tpot_p95_ms":12.105919975430762,"tpot_p99_ms":12.135692873646816,"e2e_p50_ms":25057.22738750046,"e2e_p95_ms":25164.101334703446,"e2e_p99_ms":25175.635717739642,"itl_p50_ms":11.723723000613973,"itl_p95_ms":12.08025178639218,"itl_p99_ms":12.76948868646286,"validation_errors":[],"timestamp":"2026-07-16T19:26:10+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":2048,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i16384_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i16384_o2048_a1.log"} +{"status":"COMPLETED","completed":160,"failed":0,"duration_s":262.4147839480138,"request_tps":0.6097217450663798,"input_tps":9989.681071167566,"output_tps":1248.7101338959458,"total_tps":11238.391205063512,"mean_input_tokens":16384.0,"mean_output_tokens":2048.0,"ttft_p50_ms":2030.8082264964469,"ttft_p95_ms":23697.570598886523,"ttft_p99_ms":24982.915203587618,"tpot_p50_ms":23.179559257443874,"tpot_p95_ms":31.551512422186658,"tpot_p99_ms":33.554478466176256,"e2e_p50_ms":52189.317357493564,"e2e_p95_ms":86702.39336090744,"e2e_p99_ms":87993.08046271733,"itl_p50_ms":15.3286209970247,"itl_p95_ms":15.942659782012925,"itl_p99_ms":610.8087138849077,"validation_errors":[],"timestamp":"2026-07-16T19:31:18+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":2048,"concurrency":32,"num_prompts":160,"warmup_requests":32,"attempt":1,"gain_pct":-4.501304,"plateau_streak":1,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c32_i16384_o2048_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c32_i16384_o2048_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":467.49062519401195,"request_tps":0.17112642626106017,"input_tps":11214.94147144484,"output_tps":21.904182561415702,"total_tps":11236.845654006254,"mean_input_tokens":65536.0,"mean_output_tokens":128.0,"ttft_p50_ms":28092.216243006987,"ttft_p95_ms":65609.9809892679,"ttft_p99_ms":84234.906140567,"tpot_p50_ms":529.0584835078563,"tpot_p95_ms":689.1646421382568,"tpot_p99_ms":694.2038618913791,"e2e_p50_ms":94517.5517155003,"e2e_p95_ms":132926.59808922032,"e2e_p99_ms":151589.5588841254,"itl_p50_ms":700.4347535112174,"itl_p95_ms":837.8521792372339,"itl_p99_ms":848.7232003509416,"validation_errors":[],"timestamp":"2026-07-16T19:39:53+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i65536_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i65536_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":235.13304557901574,"request_tps":0.17011645428867683,"input_tps":11148.751948262725,"output_tps":21.774906148950635,"total_tps":11170.526854411675,"mean_input_tokens":65536.0,"mean_output_tokens":128.0,"ttft_p50_ms":24172.63857051148,"ttft_p95_ms":35960.15721101285,"ttft_p99_ms":39592.695025814464,"tpot_p50_ms":166.83349701175814,"tpot_p95_ms":328.55576824153474,"tpot_p99_ms":329.2510806872486,"e2e_p50_ms":48098.485034002806,"e2e_p95_ms":51191.40079334465,"e2e_p99_ms":56645.06894614664,"itl_p50_ms":9.62428402272053,"itl_p95_ms":816.3079831079814,"itl_p99_ms":889.5722046121956,"validation_errors":[],"timestamp":"2026-07-16T19:44:34+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i65536_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i65536_o128_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":29.374553603003733,"request_tps":0.170215352633945,"input_tps":11155.23335021822,"output_tps":21.78756513714496,"total_tps":11177.020915355364,"mean_input_tokens":65536.0,"mean_output_tokens":128.0,"ttft_p50_ms":6167.725449980935,"ttft_p95_ms":6247.644395998213,"ttft_p99_ms":6256.791441601235,"tpot_p50_ms":6.4021568109673685,"tpot_p95_ms":6.590872122924481,"tpot_p99_ms":6.591545397903419,"e2e_p50_ms":7004.3386799807195,"e2e_p95_ms":7057.308506994741,"e2e_p99_ms":7065.896558194654,"itl_p50_ms":6.835169973783195,"itl_p95_ms":6.938054598867893,"itl_p99_ms":7.164961716625845,"validation_errors":[],"timestamp":"2026-07-16T19:45:48+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":128,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i65536_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i65536_o128_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":452.2059691059985,"request_tps":0.17691053516643815,"input_tps":11594.00883266769,"output_tps":45.289097002608166,"total_tps":11639.2979296703,"mean_input_tokens":65536.0,"mean_output_tokens":256.0,"ttft_p50_ms":27964.375940995524,"ttft_p95_ms":42944.88978464359,"ttft_p99_ms":61584.43124415524,"tpot_p50_ms":264.55735394904644,"tpot_p95_ms":328.878599924461,"tpot_p99_ms":350.21582279077904,"e2e_p50_ms":95521.30952950392,"e2e_p95_ms":106008.97152810794,"e2e_p99_ms":121054.057937187,"itl_p50_ms":12.545981488074176,"itl_p95_ms":818.9031008194432,"itl_p99_ms":841.0442970949224,"validation_errors":[],"timestamp":"2026-07-16T19:53:58+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":256,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i65536_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i65536_o256_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":241.68815920100315,"request_tps":0.1655025224745639,"input_tps":10846.37331289302,"output_tps":42.36864575348836,"total_tps":10888.741958646508,"mean_input_tokens":65536.0,"mean_output_tokens":256.0,"ttft_p50_ms":24228.64825501165,"ttft_p95_ms":35330.018459535495,"ttft_p99_ms":39612.372868892846,"tpot_p50_ms":88.08975900584996,"tpot_p95_ms":168.72122898116345,"tpot_p99_ms":169.04595058763002,"e2e_p50_ms":49360.89754800196,"e2e_p95_ms":52881.204059695294,"e2e_p99_ms":57884.091986017,"itl_p50_ms":9.613202011678368,"itl_p95_ms":762.7214051928601,"itl_p99_ms":843.184829771053,"validation_errors":[],"timestamp":"2026-07-16T19:58:43+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":256,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i65536_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i65536_o256_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":33.741494465008145,"request_tps":0.14818549324142372,"input_tps":9711.484485069945,"output_tps":37.93548626980447,"total_tps":9749.419971339748,"mean_input_tokens":65536.0,"mean_output_tokens":256.0,"ttft_p50_ms":6187.758629996097,"ttft_p95_ms":6245.184306602459,"ttft_p99_ms":6254.763505321462,"tpot_p50_ms":6.614399407921797,"tpot_p95_ms":6.6246079560597515,"tpot_p99_ms":6.626020995101105,"e2e_p50_ms":7877.484064985765,"e2e_p95_ms":7930.040321213892,"e2e_p99_ms":7939.256353051169,"itl_p50_ms":6.8363494938239455,"itl_p95_ms":6.942608255485538,"itl_p99_ms":7.0092535417643385,"validation_errors":[],"timestamp":"2026-07-16T20:00:05+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":256,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i65536_o256_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i65536_o256_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":468.2541085879784,"request_tps":0.1708474064247727,"input_tps":11196.655627453903,"output_tps":87.47387208948362,"total_tps":11284.129499543387,"mean_input_tokens":65536.0,"mean_output_tokens":512.0,"ttft_p50_ms":28707.305190488114,"ttft_p95_ms":42941.14238748442,"ttft_p99_ms":61583.86625094918,"tpot_p50_ms":138.2095831995791,"tpot_p95_ms":170.5406067458202,"tpot_p99_ms":181.07553997671428,"e2e_p50_ms":98761.97971450165,"e2e_p95_ms":112307.20164615194,"e2e_p99_ms":130892.13807505231,"itl_p50_ms":12.49102050496731,"itl_p95_ms":781.4320894540288,"itl_p99_ms":836.5383420526632,"validation_errors":[],"timestamp":"2026-07-16T20:08:36+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":512,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i65536_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i65536_o512_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":253.59146945900284,"request_tps":0.15773401244660815,"input_tps":10337.256239700911,"output_tps":80.75981437266337,"total_tps":10418.016054073574,"mean_input_tokens":65536.0,"mean_output_tokens":512.0,"ttft_p50_ms":18460.19749098923,"ttft_p95_ms":35239.79678586038,"ttft_p99_ms":39810.581101148564,"tpot_p50_ms":60.03944050394105,"tpot_p95_ms":88.83629755211601,"tpot_p99_ms":89.054011180707,"e2e_p50_ms":51667.23371298576,"e2e_p95_ms":64968.29380615088,"e2e_p99_ms":74357.24756326554,"itl_p50_ms":9.60926100378856,"itl_p95_ms":689.5956993947038,"itl_p99_ms":821.5999770630151,"validation_errors":[],"timestamp":"2026-07-16T20:13:34+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":512,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i65536_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i65536_o512_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":42.45334394200472,"request_tps":0.11777635247839305,"input_tps":7718.591036023967,"output_tps":60.30149246893724,"total_tps":7778.892528492904,"mean_input_tokens":65536.0,"mean_output_tokens":512.0,"ttft_p50_ms":6204.497470986098,"ttft_p95_ms":6222.823049785802,"ttft_p99_ms":6224.350896343822,"tpot_p50_ms":6.734576565566621,"tpot_p95_ms":6.741488577286328,"tpot_p99_ms":6.742628805848284,"e2e_p50_ms":9645.783593994565,"e2e_p95_ms":9660.80251399544,"e2e_p99_ms":9663.666046793805,"itl_p50_ms":6.839831999968737,"itl_p95_ms":6.9495914547587745,"itl_p99_ms":7.086883967567698,"validation_errors":[],"timestamp":"2026-07-16T20:14:58+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":512,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i65536_o512_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i65536_o512_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":500.2035044890072,"request_tps":0.15993490505774363,"input_tps":10481.493937864287,"output_tps":163.77334277912948,"total_tps":10645.267280643417,"mean_input_tokens":65536.0,"mean_output_tokens":1024.0,"ttft_p50_ms":30243.93685850373,"ttft_p95_ms":42969.91652948925,"ttft_p99_ms":61614.389155033525,"tpot_p50_ms":73.36121226490721,"tpot_p95_ms":91.33674583832028,"tpot_p99_ms":96.64777304192485,"e2e_p50_ms":105178.19468899688,"e2e_p95_ms":113684.78196981303,"e2e_p99_ms":132258.0071013729,"itl_p50_ms":12.48324201151263,"itl_p95_ms":713.5221524600638,"itl_p99_ms":825.3858603240224,"validation_errors":[],"timestamp":"2026-07-16T20:24:07+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":1024,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i65536_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i65536_o1024_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":278.18836335701053,"request_tps":0.14378746658308766,"input_tps":9423.255409989233,"output_tps":147.23836578108177,"total_tps":9570.493775770316,"mean_input_tokens":65536.0,"mean_output_tokens":1024.0,"ttft_p50_ms":24218.15920100198,"ttft_p95_ms":37847.32255409762,"ttft_p99_ms":40382.343416879594,"tpot_p50_ms":29.127336590921143,"tpot_p95_ms":49.198489598973694,"tpot_p99_ms":49.27219220356028,"e2e_p50_ms":56596.45105450181,"e2e_p95_ms":56909.13160444033,"e2e_p99_ms":62302.39748750086,"itl_p50_ms":9.62444901233539,"itl_p95_ms":9.987523194286041,"itl_p99_ms":776.7235341103515,"validation_errors":[],"timestamp":"2026-07-16T20:29:32+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":1024,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i65536_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i65536_o1024_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":60.06458672598819,"request_tps":0.08324372600463838,"input_tps":5455.460827439981,"output_tps":85.2415754287497,"total_tps":5540.70240286873,"mean_input_tokens":65536.0,"mean_output_tokens":1024.0,"ttft_p50_ms":6200.891699991189,"ttft_p95_ms":6272.143243986648,"ttft_p99_ms":6285.5985119822435,"tpot_p50_ms":6.7929081173096115,"tpot_p95_ms":6.796755083673428,"tpot_p99_ms":6.797365722498417,"e2e_p50_ms":13150.848732999293,"e2e_p95_ms":13210.16350100399,"e2e_p99_ms":13220.45055940398,"itl_p50_ms":6.836396001745015,"itl_p95_ms":6.958074710564688,"itl_p99_ms":7.039473712211475,"validation_errors":[],"timestamp":"2026-07-16T20:31:14+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":1024,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i65536_o1024_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i65536_o1024_a1.log"} +{"status":"COMPLETED","completed":80,"failed":0,"duration_s":1061.9543217770115,"request_tps":0.07533280703272881,"input_tps":9874.02168339383,"output_tps":9.642599300189287,"total_tps":9883.664282694019,"mean_input_tokens":131072.0,"mean_output_tokens":128.0,"ttft_p50_ms":109194.33135900181,"ttft_p95_ms":129901.68803883859,"ttft_p99_ms":173369.75827682696,"tpot_p50_ms":860.7348704526814,"tpot_p95_ms":863.007278436844,"tpot_p99_ms":877.5824045137604,"e2e_p50_ms":218323.68925350602,"e2e_p95_ms":239327.92834405843,"e2e_p99_ms":282901.98067860294,"itl_p50_ms":857.2086700005457,"itl_p95_ms":1084.6890412460198,"itl_p99_ms":1101.4063146195258,"validation_errors":[],"timestamp":"2026-07-16T20:49:47+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":131072,"osl":128,"concurrency":16,"num_prompts":80,"warmup_requests":16,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c16_i131072_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c16_i131072_o128_a1.log"} +{"status":"COMPLETED","completed":40,"failed":0,"duration_s":542.2442093600112,"request_tps":0.07376750052750287,"input_tps":9668.853829140857,"output_tps":9.442240067520368,"total_tps":9678.296069208376,"mean_input_tokens":131072.0,"mean_output_tokens":128.0,"ttft_p50_ms":56039.164249494206,"ttft_p95_ms":72600.3135224935,"ttft_p99_ms":92221.81805258355,"tpot_p50_ms":387.13930614164747,"tpot_p95_ms":753.8479004563203,"tpot_p99_ms":754.9241562044031,"e2e_p50_ms":110909.34608949465,"e2e_p95_ms":112789.03825180455,"e2e_p99_ms":134558.94819398702,"itl_p50_ms":660.2215159800835,"itl_p95_ms":1072.356289398158,"itl_p99_ms":1163.3652878389696,"validation_errors":[],"timestamp":"2026-07-16T20:59:41+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":131072,"osl":128,"concurrency":8,"num_prompts":40,"warmup_requests":8,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c8_i131072_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c8_i131072_o128_a1.log"} +{"status":"COMPLETED","completed":5,"failed":0,"duration_s":62.04160683800001,"request_tps":0.08059107838802039,"input_tps":10563.233826474609,"output_tps":10.31565803366661,"total_tps":10573.549484508274,"mean_input_tokens":131072.0,"mean_output_tokens":128.0,"ttft_p50_ms":14315.946837014053,"ttft_p95_ms":14424.979343393352,"ttft_p99_ms":14444.436821474228,"tpot_p50_ms":6.151024771691117,"tpot_p95_ms":6.183124977891795,"tpot_p99_ms":6.187243395545484,"e2e_p50_ms":15091.257207997842,"e2e_p95_ms":15205.586195399519,"e2e_p99_ms":15225.502308679279,"itl_p50_ms":6.912430500960909,"itl_p95_ms":7.0415690206573345,"itl_p99_ms":7.134565238084179,"validation_errors":[],"timestamp":"2026-07-16T21:01:36+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":131072,"osl":128,"concurrency":1,"num_prompts":5,"warmup_requests":1,"attempt":1,"gain_pct":null,"plateau_streak":0,"error_type":"","raw_file":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/raw_outputs/vllm_adaptive_c1_i131072_o128_a1.jsonl","detail_log":"/data1/yy/sskj/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/logs/vllm_c1_i131072_o128_a1.log"} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_shapes.jsonl b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_shapes.jsonl index 6850b0f..9dbd87f 100644 --- a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_shapes.jsonl +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_shapes.jsonl @@ -25,3 +25,47 @@ {"timestamp":"2026-07-16T10:40:50+00:00","engine":"vllm","tp":2,"dp":4,"mark":"Y","isl":262144,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":13879.801995316056,"last_total_tps":6075.0183795289} {"timestamp":"2026-07-16T11:28:54+00:00","engine":"vllm","tp":2,"dp":4,"mark":"Y","isl":262144,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":13481.264910170603,"last_total_tps":5891.6134230831085} {"timestamp":"2026-07-16T12:39:42+00:00","engine":"vllm","tp":2,"dp":4,"mark":"Y","isl":524288,"osl":128,"status":"FAILED","stop_reason":"BENCHMARK_FAILED","tested_points":1,"search_cap":512,"max_successful_concurrency":0,"saturation_concurrency":null,"stop_probe_concurrency":16,"best_tps_concurrency":null,"best_total_tps":null,"last_total_tps":null} +{"timestamp":"2026-07-16T13:14:35+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":32,"best_total_tps":1928.0659392758148,"last_total_tps":1928.0659392758148} +{"timestamp":"2026-07-16T13:17:38+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":8,"best_total_tps":2768.1737168097607,"last_total_tps":2768.1737168097607} +{"timestamp":"2026-07-16T13:27:33+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":6,"search_cap":512,"max_successful_concurrency":96,"saturation_concurrency":null,"stop_probe_concurrency":96,"best_tps_concurrency":96,"best_total_tps":9824.100767069867,"last_total_tps":9824.100767069867} +{"timestamp":"2026-07-16T13:45:29+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":7,"search_cap":512,"max_successful_concurrency":112,"saturation_concurrency":null,"stop_probe_concurrency":112,"best_tps_concurrency":112,"best_total_tps":7985.1662368534835,"last_total_tps":7985.1662368534835} +{"timestamp":"2026-07-16T14:05:45+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":2048,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":5,"search_cap":512,"max_successful_concurrency":80,"saturation_concurrency":null,"stop_probe_concurrency":80,"best_tps_concurrency":80,"best_total_tps":4621.455132269518,"last_total_tps":4621.455132269518} +{"timestamp":"2026-07-16T14:50:01+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":1024,"osl":4096,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":6,"search_cap":512,"max_successful_concurrency":96,"saturation_concurrency":null,"stop_probe_concurrency":96,"best_tps_concurrency":96,"best_total_tps":4822.584204271116,"last_total_tps":4822.584204271116} +{"timestamp":"2026-07-16T14:52:41+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":8,"best_total_tps":12691.712913192754,"last_total_tps":12691.712913192754} +{"timestamp":"2026-07-16T14:55:12+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":32,"best_total_tps":17822.829023404527,"last_total_tps":17822.829023404527} +{"timestamp":"2026-07-16T15:00:08+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":48,"saturation_concurrency":null,"stop_probe_concurrency":48,"best_tps_concurrency":48,"best_total_tps":14064.895332699014,"last_total_tps":14064.895332699014} +{"timestamp":"2026-07-16T15:06:46+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":48,"saturation_concurrency":null,"stop_probe_concurrency":48,"best_tps_concurrency":48,"best_total_tps":11482.884078475714,"last_total_tps":11482.884078475714} +{"timestamp":"2026-07-16T15:22:09+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":2048,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":4,"search_cap":512,"max_successful_concurrency":64,"saturation_concurrency":null,"stop_probe_concurrency":64,"best_tps_concurrency":64,"best_total_tps":8161.264785058724,"last_total_tps":8161.264785058724} +{"timestamp":"2026-07-16T15:34:10+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":4096,"osl":4096,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":32,"best_total_tps":3629.0218039514316,"last_total_tps":3629.0218039514316} +{"timestamp":"2026-07-16T15:36:59+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":8,"best_total_tps":31173.237551508326,"last_total_tps":31173.237551508326} +{"timestamp":"2026-07-16T15:39:33+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":16,"best_total_tps":27719.57133134522,"last_total_tps":26528.84610875391} +{"timestamp":"2026-07-16T15:42:25+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":16,"best_total_tps":24136.4598371369,"last_total_tps":20630.058591354147} +{"timestamp":"2026-07-16T15:48:05+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":16,"best_total_tps":17026.620956630413,"last_total_tps":16874.05385755264} +{"timestamp":"2026-07-16T15:55:10+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":16384,"osl":2048,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":32,"best_total_tps":14527.480180568706,"last_total_tps":14527.480180568706} +{"timestamp":"2026-07-16T16:07:06+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":15790.495048183519,"last_total_tps":8442.615514866722} +{"timestamp":"2026-07-16T16:19:29+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":15474.072939048607,"last_total_tps":7397.5580917508105} +{"timestamp":"2026-07-16T16:32:37+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":14467.082373173163,"last_total_tps":5964.714088744037} +{"timestamp":"2026-07-16T16:46:57+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":65536,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":13621.063853724581,"last_total_tps":4329.257770068971} +{"timestamp":"2026-07-16T17:10:42+00:00","engine":"vllm","tp":4,"dp":2,"mark":"Y","isl":131072,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_GROUP_SKIP_THRESHOLD","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":8,"best_total_tps":13775.839598604127,"last_total_tps":7982.424026887654} +{"timestamp":"2026-07-16T17:24:40+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":1771.461822081285,"last_total_tps":1250.0250146108165} +{"timestamp":"2026-07-16T17:27:58+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":8,"best_total_tps":2142.3842280060862,"last_total_tps":720.8602718842554} +{"timestamp":"2026-07-16T17:32:54+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":48,"saturation_concurrency":null,"stop_probe_concurrency":48,"best_tps_concurrency":32,"best_total_tps":4817.613398959913,"last_total_tps":4331.306660759834} +{"timestamp":"2026-07-16T17:44:07+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":5,"search_cap":512,"max_successful_concurrency":80,"saturation_concurrency":null,"stop_probe_concurrency":80,"best_tps_concurrency":80,"best_total_tps":6231.279678122503,"last_total_tps":6231.279678122503} +{"timestamp":"2026-07-16T17:58:42+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":2048,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":4,"search_cap":512,"max_successful_concurrency":64,"saturation_concurrency":null,"stop_probe_concurrency":64,"best_tps_concurrency":64,"best_total_tps":4249.03535847688,"last_total_tps":4249.03535847688} +{"timestamp":"2026-07-16T18:33:07+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":1024,"osl":4096,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":5,"search_cap":512,"max_successful_concurrency":80,"saturation_concurrency":null,"stop_probe_concurrency":80,"best_tps_concurrency":80,"best_total_tps":4082.095146403349,"last_total_tps":4082.095146403349} +{"timestamp":"2026-07-16T18:35:04+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":8,"best_total_tps":20803.05786963908,"last_total_tps":20803.05786963908} +{"timestamp":"2026-07-16T18:37:16+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":16,"best_total_tps":19642.58502143222,"last_total_tps":15091.895727591496} +{"timestamp":"2026-07-16T18:41:51+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":48,"saturation_concurrency":null,"stop_probe_concurrency":48,"best_tps_concurrency":32,"best_total_tps":16790.78056362789,"last_total_tps":14485.55087923292} +{"timestamp":"2026-07-16T18:45:53+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":32,"best_total_tps":7915.097557814089,"last_total_tps":7915.097557814089} +{"timestamp":"2026-07-16T18:56:11+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":2048,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":48,"saturation_concurrency":null,"stop_probe_concurrency":48,"best_tps_concurrency":48,"best_total_tps":7130.506489441016,"last_total_tps":7130.506489441016} +{"timestamp":"2026-07-16T19:07:02+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":4096,"osl":4096,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":32,"best_total_tps":3935.4715147846564,"last_total_tps":3935.4715147846564} +{"timestamp":"2026-07-16T19:09:55+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":8,"best_total_tps":74710.91951665143,"last_total_tps":74710.91951665143} +{"timestamp":"2026-07-16T19:13:39+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":16,"best_total_tps":65665.0899763922,"last_total_tps":21252.567939828114} +{"timestamp":"2026-07-16T19:19:23+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":10378.673726315716,"last_total_tps":4652.484093418} +{"timestamp":"2026-07-16T19:23:23+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_BACKOFF_FOUND","tested_points":2,"search_cap":512,"max_successful_concurrency":8,"saturation_concurrency":null,"stop_probe_concurrency":8,"best_tps_concurrency":8,"best_total_tps":13756.653148310354,"last_total_tps":13756.653148310354} +{"timestamp":"2026-07-16T19:31:18+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":16384,"osl":2048,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":2,"search_cap":512,"max_successful_concurrency":32,"saturation_concurrency":null,"stop_probe_concurrency":32,"best_tps_concurrency":16,"best_total_tps":11768.109544317756,"last_total_tps":11238.391205063512} +{"timestamp":"2026-07-16T19:45:48+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":11236.845654006254,"last_total_tps":11177.020915355364} +{"timestamp":"2026-07-16T20:00:05+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":256,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":11639.2979296703,"last_total_tps":9749.419971339748} +{"timestamp":"2026-07-16T20:14:59+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":512,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":11284.129499543387,"last_total_tps":7778.892528492904} +{"timestamp":"2026-07-16T20:31:14+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":65536,"osl":1024,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_SLO_EXCEEDED","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":16,"best_total_tps":10645.267280643417,"last_total_tps":5540.70240286873} +{"timestamp":"2026-07-16T21:01:36+00:00","engine":"vllm","tp":8,"dp":1,"mark":"Y","isl":131072,"osl":128,"status":"TTFT_BOUNDARY","stop_reason":"TTFT_GROUP_SKIP_THRESHOLD","tested_points":3,"search_cap":512,"max_successful_concurrency":1,"saturation_concurrency":null,"stop_probe_concurrency":1,"best_tps_concurrency":1,"best_total_tps":10573.549484508274,"last_total_tps":10573.549484508274} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.jsonl b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.jsonl new file mode 100644 index 0000000..298bbaf --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.jsonl @@ -0,0 +1,71 @@ +{"timestamp": "2026-07-16T05:31:54+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 1024, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 3646.9396491060006, "last_total_tps": 3646.9396491060006} +{"timestamp": "2026-07-16T05:35:22+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 1024, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 16, "best_total_tps": 4016.3582354971636, "last_total_tps": 3510.8603471897136} +{"timestamp": "2026-07-16T05:37:07+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 1024, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 1933.4323295182799, "last_total_tps": 1933.4323295182799} +{"timestamp": "2026-07-16T05:50:06+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 1024, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 5, "search_cap": 512, "max_successful_concurrency": 80, "saturation_concurrency": null, "stop_probe_concurrency": 80, "best_tps_concurrency": 80, "best_total_tps": 5711.313811148789, "last_total_tps": 5711.313811148789} +{"timestamp": "2026-07-16T06:21:25+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 1024, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 7, "search_cap": 512, "max_successful_concurrency": 112, "saturation_concurrency": null, "stop_probe_concurrency": 112, "best_tps_concurrency": 112, "best_total_tps": 6326.121211706281, "last_total_tps": 6326.121211706281} +{"timestamp": "2026-07-16T07:08:21+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 1024, "osl": 4096, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 6, "search_cap": 512, "max_successful_concurrency": 96, "saturation_concurrency": null, "stop_probe_concurrency": 96, "best_tps_concurrency": 96, "best_total_tps": 4685.666489411614, "last_total_tps": 4685.666489411614} +{"timestamp": "2026-07-16T07:09:51+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 4096, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 7626.584143741924, "last_total_tps": 7626.584143741924} +{"timestamp": "2026-07-16T07:12:56+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 4096, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 16, "best_total_tps": 10703.552130441196, "last_total_tps": 10508.443270196269} +{"timestamp": "2026-07-16T07:18:17+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 4096, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 14258.166155126417, "last_total_tps": 14258.166155126417} +{"timestamp": "2026-07-16T07:25:36+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 4096, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 10711.105514976523, "last_total_tps": 10711.105514976523} +{"timestamp": "2026-07-16T07:37:17+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 4096, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 7033.604800110782, "last_total_tps": 7033.604800110782} +{"timestamp": "2026-07-16T07:57:45+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 4096, "osl": 4096, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 4885.314271518819, "last_total_tps": 4885.314271518819} +{"timestamp": "2026-07-16T07:59:21+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 16384, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 23587.87688648832, "last_total_tps": 23587.87688648832} +{"timestamp": "2026-07-16T08:01:06+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 16384, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 20694.07882326918, "last_total_tps": 20694.07882326918} +{"timestamp": "2026-07-16T08:03:06+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 16384, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 18665.914037481325, "last_total_tps": 18665.914037481325} +{"timestamp": "2026-07-16T08:05:39+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 16384, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 12488.958537642658, "last_total_tps": 12488.958537642658} +{"timestamp": "2026-07-16T08:09:28+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 16384, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 7807.1888546940745, "last_total_tps": 7807.1888546940745} +{"timestamp": "2026-07-16T08:14:46+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 65536, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 19964.966947119654, "last_total_tps": 19964.966947119654} +{"timestamp": "2026-07-16T08:21:04+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 65536, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 16516.087929598416, "last_total_tps": 16516.087929598416} +{"timestamp": "2026-07-16T08:27:37+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 65536, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 15640.847692576317, "last_total_tps": 15640.847692576317} +{"timestamp": "2026-07-16T08:34:46+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 65536, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 14124.215996887702, "last_total_tps": 14124.215996887702} +{"timestamp": "2026-07-16T08:46:49+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 131072, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 16, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": 16, "best_total_tps": 16066.556870612381, "last_total_tps": 16066.556870612381} +{"timestamp": "2026-07-16T09:25:11+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 131072, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 8, "best_total_tps": 15402.074234300384, "last_total_tps": 6752.029991153788} +{"timestamp": "2026-07-16T09:47:57+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 131072, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 15760.422764455148, "last_total_tps": 5974.45197601345} +{"timestamp": "2026-07-16T10:40:50+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 262144, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 13879.801995316056, "last_total_tps": 6075.0183795289} +{"timestamp": "2026-07-16T11:28:54+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 262144, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 13481.264910170603, "last_total_tps": 5891.6134230831085} +{"timestamp": "2026-07-16T12:39:42+00:00", "engine": "vllm", "tp": 2, "dp": 4, "mark": "Y", "isl": 524288, "osl": 128, "status": "FAILED", "stop_reason": "BENCHMARK_FAILED", "tested_points": 1, "search_cap": 512, "max_successful_concurrency": 0, "saturation_concurrency": null, "stop_probe_concurrency": 16, "best_tps_concurrency": null, "best_total_tps": null, "last_total_tps": null} +{"timestamp": "2026-07-16T13:14:35+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 1024, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 32, "best_total_tps": 1928.0659392758148, "last_total_tps": 1928.0659392758148} +{"timestamp": "2026-07-16T13:17:38+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 1024, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 8, "best_total_tps": 2768.1737168097607, "last_total_tps": 2768.1737168097607} +{"timestamp": "2026-07-16T13:27:33+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 1024, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 6, "search_cap": 512, "max_successful_concurrency": 96, "saturation_concurrency": null, "stop_probe_concurrency": 96, "best_tps_concurrency": 96, "best_total_tps": 9824.100767069867, "last_total_tps": 9824.100767069867} +{"timestamp": "2026-07-16T13:45:29+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 1024, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 7, "search_cap": 512, "max_successful_concurrency": 112, "saturation_concurrency": null, "stop_probe_concurrency": 112, "best_tps_concurrency": 112, "best_total_tps": 7985.1662368534835, "last_total_tps": 7985.1662368534835} +{"timestamp": "2026-07-16T14:05:45+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 1024, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 5, "search_cap": 512, "max_successful_concurrency": 80, "saturation_concurrency": null, "stop_probe_concurrency": 80, "best_tps_concurrency": 80, "best_total_tps": 4621.455132269518, "last_total_tps": 4621.455132269518} +{"timestamp": "2026-07-16T14:50:01+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 1024, "osl": 4096, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 6, "search_cap": 512, "max_successful_concurrency": 96, "saturation_concurrency": null, "stop_probe_concurrency": 96, "best_tps_concurrency": 96, "best_total_tps": 4822.584204271116, "last_total_tps": 4822.584204271116} +{"timestamp": "2026-07-16T14:52:41+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 4096, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 8, "best_total_tps": 12691.712913192754, "last_total_tps": 12691.712913192754} +{"timestamp": "2026-07-16T14:55:12+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 4096, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 32, "best_total_tps": 17822.829023404527, "last_total_tps": 17822.829023404527} +{"timestamp": "2026-07-16T15:00:08+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 4096, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 14064.895332699014, "last_total_tps": 14064.895332699014} +{"timestamp": "2026-07-16T15:06:46+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 4096, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 11482.884078475714, "last_total_tps": 11482.884078475714} +{"timestamp": "2026-07-16T15:22:09+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 4096, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 4, "search_cap": 512, "max_successful_concurrency": 64, "saturation_concurrency": null, "stop_probe_concurrency": 64, "best_tps_concurrency": 64, "best_total_tps": 8161.264785058724, "last_total_tps": 8161.264785058724} +{"timestamp": "2026-07-16T15:34:10+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 4096, "osl": 4096, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 32, "best_total_tps": 3629.0218039514316, "last_total_tps": 3629.0218039514316} +{"timestamp": "2026-07-16T15:36:59+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 16384, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 8, "best_total_tps": 31173.237551508326, "last_total_tps": 31173.237551508326} +{"timestamp": "2026-07-16T15:39:33+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 16384, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 16, "best_total_tps": 27719.57133134522, "last_total_tps": 26528.84610875391} +{"timestamp": "2026-07-16T15:42:25+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 16384, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 16, "best_total_tps": 24136.4598371369, "last_total_tps": 20630.058591354147} +{"timestamp": "2026-07-16T15:48:05+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 16384, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 16, "best_total_tps": 17026.620956630413, "last_total_tps": 16874.05385755264} +{"timestamp": "2026-07-16T15:55:10+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 16384, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 32, "best_total_tps": 14527.480180568706, "last_total_tps": 14527.480180568706} +{"timestamp": "2026-07-16T16:07:06+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 65536, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 15790.495048183519, "last_total_tps": 8442.615514866722} +{"timestamp": "2026-07-16T16:19:29+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 65536, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 15474.072939048607, "last_total_tps": 7397.5580917508105} +{"timestamp": "2026-07-16T16:32:37+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 65536, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 14467.082373173163, "last_total_tps": 5964.714088744037} +{"timestamp": "2026-07-16T16:46:57+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 65536, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 13621.063853724581, "last_total_tps": 4329.257770068971} +{"timestamp": "2026-07-16T17:10:42+00:00", "engine": "vllm", "tp": 4, "dp": 2, "mark": "Y", "isl": 131072, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_GROUP_SKIP_THRESHOLD", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 8, "best_total_tps": 13775.839598604127, "last_total_tps": 7982.424026887654} +{"timestamp": "2026-07-16T17:24:40+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 1024, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 1771.461822081285, "last_total_tps": 1250.0250146108165} +{"timestamp": "2026-07-16T17:27:58+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 1024, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 8, "best_total_tps": 2142.3842280060862, "last_total_tps": 720.8602718842554} +{"timestamp": "2026-07-16T17:32:54+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 1024, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 32, "best_total_tps": 4817.613398959913, "last_total_tps": 4331.306660759834} +{"timestamp": "2026-07-16T17:44:07+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 1024, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 5, "search_cap": 512, "max_successful_concurrency": 80, "saturation_concurrency": null, "stop_probe_concurrency": 80, "best_tps_concurrency": 80, "best_total_tps": 6231.279678122503, "last_total_tps": 6231.279678122503} +{"timestamp": "2026-07-16T17:58:42+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 1024, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 4, "search_cap": 512, "max_successful_concurrency": 64, "saturation_concurrency": null, "stop_probe_concurrency": 64, "best_tps_concurrency": 64, "best_total_tps": 4249.03535847688, "last_total_tps": 4249.03535847688} +{"timestamp": "2026-07-16T18:33:07+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 1024, "osl": 4096, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 5, "search_cap": 512, "max_successful_concurrency": 80, "saturation_concurrency": null, "stop_probe_concurrency": 80, "best_tps_concurrency": 80, "best_total_tps": 4082.095146403349, "last_total_tps": 4082.095146403349} +{"timestamp": "2026-07-16T18:35:04+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 4096, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 8, "best_total_tps": 20803.05786963908, "last_total_tps": 20803.05786963908} +{"timestamp": "2026-07-16T18:37:16+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 4096, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 16, "best_total_tps": 19642.58502143222, "last_total_tps": 15091.895727591496} +{"timestamp": "2026-07-16T18:41:51+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 4096, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 32, "best_total_tps": 16790.78056362789, "last_total_tps": 14485.55087923292} +{"timestamp": "2026-07-16T18:45:53+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 4096, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 32, "best_total_tps": 7915.097557814089, "last_total_tps": 7915.097557814089} +{"timestamp": "2026-07-16T18:56:11+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 4096, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 48, "saturation_concurrency": null, "stop_probe_concurrency": 48, "best_tps_concurrency": 48, "best_total_tps": 7130.506489441016, "last_total_tps": 7130.506489441016} +{"timestamp": "2026-07-16T19:07:02+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 4096, "osl": 4096, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 32, "best_total_tps": 3935.4715147846564, "last_total_tps": 3935.4715147846564} +{"timestamp": "2026-07-16T19:09:55+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 16384, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 8, "best_total_tps": 74710.91951665143, "last_total_tps": 74710.91951665143} +{"timestamp": "2026-07-16T19:13:39+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 16384, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 16, "best_total_tps": 65665.0899763922, "last_total_tps": 21252.567939828114} +{"timestamp": "2026-07-16T19:19:23+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 16384, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 10378.673726315716, "last_total_tps": 4652.484093418} +{"timestamp": "2026-07-16T19:23:23+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 16384, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_BACKOFF_FOUND", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 8, "saturation_concurrency": null, "stop_probe_concurrency": 8, "best_tps_concurrency": 8, "best_total_tps": 13756.653148310354, "last_total_tps": 13756.653148310354} +{"timestamp": "2026-07-16T19:31:18+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 16384, "osl": 2048, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 2, "search_cap": 512, "max_successful_concurrency": 32, "saturation_concurrency": null, "stop_probe_concurrency": 32, "best_tps_concurrency": 16, "best_total_tps": 11768.109544317756, "last_total_tps": 11238.391205063512} +{"timestamp": "2026-07-16T19:45:48+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 65536, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 11236.845654006254, "last_total_tps": 11177.020915355364} +{"timestamp": "2026-07-16T20:00:05+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 65536, "osl": 256, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 11639.2979296703, "last_total_tps": 9749.419971339748} +{"timestamp": "2026-07-16T20:14:59+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 65536, "osl": 512, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 11284.129499543387, "last_total_tps": 7778.892528492904} +{"timestamp": "2026-07-16T20:31:14+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 65536, "osl": 1024, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_SLO_EXCEEDED", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 16, "best_total_tps": 10645.267280643417, "last_total_tps": 5540.70240286873} +{"timestamp": "2026-07-16T21:01:36+00:00", "engine": "vllm", "tp": 8, "dp": 1, "mark": "Y", "isl": 131072, "osl": 128, "status": "TTFT_BOUNDARY", "stop_reason": "TTFT_GROUP_SKIP_THRESHOLD", "tested_points": 3, "search_cap": 512, "max_successful_concurrency": 1, "saturation_concurrency": null, "stop_probe_concurrency": 1, "best_tps_concurrency": 1, "best_total_tps": 10573.549484508274, "last_total_tps": 10573.549484508274} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.md b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.md new file mode 100644 index 0000000..97a1ed8 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/adaptive_summary.md @@ -0,0 +1,77 @@ +# Adaptive concurrency search summary + +| Engine | TP | DP | ISL | OSL | Stop | Saturation C | Best TPS C | Best Total TPS | Max successful C | +|---|---:|---:|---:|---:|---|---:|---:|---:|---:| +| vllm | 2 | 4 | 1024 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 3646.9396491060006 | 16 | +| vllm | 2 | 4 | 1024 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 4016.3582354971636 | 32 | +| vllm | 2 | 4 | 1024 | 512 | TTFT_SLO_EXCEEDED | None | 16 | 1933.4323295182799 | 16 | +| vllm | 2 | 4 | 1024 | 1024 | TTFT_SLO_EXCEEDED | None | 80 | 5711.313811148789 | 80 | +| vllm | 2 | 4 | 1024 | 2048 | TTFT_SLO_EXCEEDED | None | 112 | 6326.121211706281 | 112 | +| vllm | 2 | 4 | 1024 | 4096 | TTFT_SLO_EXCEEDED | None | 96 | 4685.666489411614 | 96 | +| vllm | 2 | 4 | 4096 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 7626.584143741924 | 16 | +| vllm | 2 | 4 | 4096 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 10703.552130441196 | 32 | +| vllm | 2 | 4 | 4096 | 512 | TTFT_SLO_EXCEEDED | None | 48 | 14258.166155126417 | 48 | +| vllm | 2 | 4 | 4096 | 1024 | TTFT_SLO_EXCEEDED | None | 48 | 10711.105514976523 | 48 | +| vllm | 2 | 4 | 4096 | 2048 | TTFT_SLO_EXCEEDED | None | 48 | 7033.604800110782 | 48 | +| vllm | 2 | 4 | 4096 | 4096 | TTFT_SLO_EXCEEDED | None | 48 | 4885.314271518819 | 48 | +| vllm | 2 | 4 | 16384 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 23587.87688648832 | 16 | +| vllm | 2 | 4 | 16384 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 20694.07882326918 | 16 | +| vllm | 2 | 4 | 16384 | 512 | TTFT_SLO_EXCEEDED | None | 16 | 18665.914037481325 | 16 | +| vllm | 2 | 4 | 16384 | 1024 | TTFT_SLO_EXCEEDED | None | 16 | 12488.958537642658 | 16 | +| vllm | 2 | 4 | 16384 | 2048 | TTFT_SLO_EXCEEDED | None | 16 | 7807.1888546940745 | 16 | +| vllm | 2 | 4 | 65536 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 19964.966947119654 | 16 | +| vllm | 2 | 4 | 65536 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 16516.087929598416 | 16 | +| vllm | 2 | 4 | 65536 | 512 | TTFT_SLO_EXCEEDED | None | 16 | 15640.847692576317 | 16 | +| vllm | 2 | 4 | 65536 | 1024 | TTFT_SLO_EXCEEDED | None | 16 | 14124.215996887702 | 16 | +| vllm | 2 | 4 | 131072 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 16066.556870612381 | 16 | +| vllm | 2 | 4 | 131072 | 256 | TTFT_SLO_EXCEEDED | None | 8 | 15402.074234300384 | 1 | +| vllm | 2 | 4 | 131072 | 512 | TTFT_SLO_EXCEEDED | None | 16 | 15760.422764455148 | 1 | +| vllm | 2 | 4 | 262144 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 13879.801995316056 | 1 | +| vllm | 2 | 4 | 262144 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 13481.264910170603 | 1 | +| vllm | 2 | 4 | 524288 | 128 | BENCHMARK_FAILED | None | None | None | 0 | +| vllm | 4 | 2 | 1024 | 128 | TTFT_SLO_EXCEEDED | None | 32 | 1928.0659392758148 | 32 | +| vllm | 4 | 2 | 1024 | 256 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 2768.1737168097607 | 8 | +| vllm | 4 | 2 | 1024 | 512 | TTFT_SLO_EXCEEDED | None | 96 | 9824.100767069867 | 96 | +| vllm | 4 | 2 | 1024 | 1024 | TTFT_SLO_EXCEEDED | None | 112 | 7985.1662368534835 | 112 | +| vllm | 4 | 2 | 1024 | 2048 | TTFT_SLO_EXCEEDED | None | 80 | 4621.455132269518 | 80 | +| vllm | 4 | 2 | 1024 | 4096 | TTFT_SLO_EXCEEDED | None | 96 | 4822.584204271116 | 96 | +| vllm | 4 | 2 | 4096 | 128 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 12691.712913192754 | 8 | +| vllm | 4 | 2 | 4096 | 256 | TTFT_SLO_EXCEEDED | None | 32 | 17822.829023404527 | 32 | +| vllm | 4 | 2 | 4096 | 512 | TTFT_SLO_EXCEEDED | None | 48 | 14064.895332699014 | 48 | +| vllm | 4 | 2 | 4096 | 1024 | TTFT_SLO_EXCEEDED | None | 48 | 11482.884078475714 | 48 | +| vllm | 4 | 2 | 4096 | 2048 | TTFT_SLO_EXCEEDED | None | 64 | 8161.264785058724 | 64 | +| vllm | 4 | 2 | 4096 | 4096 | TTFT_SLO_EXCEEDED | None | 32 | 3629.0218039514316 | 32 | +| vllm | 4 | 2 | 16384 | 128 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 31173.237551508326 | 8 | +| vllm | 4 | 2 | 16384 | 256 | TTFT_SLO_BACKOFF_FOUND | None | 16 | 27719.57133134522 | 8 | +| vllm | 4 | 2 | 16384 | 512 | TTFT_SLO_BACKOFF_FOUND | None | 16 | 24136.4598371369 | 8 | +| vllm | 4 | 2 | 16384 | 1024 | TTFT_SLO_EXCEEDED | None | 16 | 17026.620956630413 | 32 | +| vllm | 4 | 2 | 16384 | 2048 | TTFT_SLO_EXCEEDED | None | 32 | 14527.480180568706 | 32 | +| vllm | 4 | 2 | 65536 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 15790.495048183519 | 1 | +| vllm | 4 | 2 | 65536 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 15474.072939048607 | 1 | +| vllm | 4 | 2 | 65536 | 512 | TTFT_SLO_EXCEEDED | None | 16 | 14467.082373173163 | 1 | +| vllm | 4 | 2 | 65536 | 1024 | TTFT_SLO_EXCEEDED | None | 16 | 13621.063853724581 | 1 | +| vllm | 4 | 2 | 131072 | 128 | TTFT_GROUP_SKIP_THRESHOLD | None | 8 | 13775.839598604127 | 1 | +| vllm | 8 | 1 | 1024 | 128 | TTFT_SLO_BACKOFF_FOUND | None | 16 | 1771.461822081285 | 1 | +| vllm | 8 | 1 | 1024 | 256 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 2142.3842280060862 | 1 | +| vllm | 8 | 1 | 1024 | 512 | TTFT_SLO_EXCEEDED | None | 32 | 4817.613398959913 | 48 | +| vllm | 8 | 1 | 1024 | 1024 | TTFT_SLO_EXCEEDED | None | 80 | 6231.279678122503 | 80 | +| vllm | 8 | 1 | 1024 | 2048 | TTFT_SLO_EXCEEDED | None | 64 | 4249.03535847688 | 64 | +| vllm | 8 | 1 | 1024 | 4096 | TTFT_SLO_EXCEEDED | None | 80 | 4082.095146403349 | 80 | +| vllm | 8 | 1 | 4096 | 128 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 20803.05786963908 | 8 | +| vllm | 8 | 1 | 4096 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 19642.58502143222 | 32 | +| vllm | 8 | 1 | 4096 | 512 | TTFT_SLO_EXCEEDED | None | 32 | 16790.78056362789 | 48 | +| vllm | 8 | 1 | 4096 | 1024 | TTFT_SLO_EXCEEDED | None | 32 | 7915.097557814089 | 32 | +| vllm | 8 | 1 | 4096 | 2048 | TTFT_SLO_EXCEEDED | None | 48 | 7130.506489441016 | 48 | +| vllm | 8 | 1 | 4096 | 4096 | TTFT_SLO_EXCEEDED | None | 32 | 3935.4715147846564 | 32 | +| vllm | 8 | 1 | 16384 | 128 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 74710.91951665143 | 8 | +| vllm | 8 | 1 | 16384 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 65665.0899763922 | 32 | +| vllm | 8 | 1 | 16384 | 512 | TTFT_SLO_BACKOFF_FOUND | None | 16 | 10378.673726315716 | 1 | +| vllm | 8 | 1 | 16384 | 1024 | TTFT_SLO_BACKOFF_FOUND | None | 8 | 13756.653148310354 | 8 | +| vllm | 8 | 1 | 16384 | 2048 | TTFT_SLO_EXCEEDED | None | 16 | 11768.109544317756 | 32 | +| vllm | 8 | 1 | 65536 | 128 | TTFT_SLO_EXCEEDED | None | 16 | 11236.845654006254 | 1 | +| vllm | 8 | 1 | 65536 | 256 | TTFT_SLO_EXCEEDED | None | 16 | 11639.2979296703 | 1 | +| vllm | 8 | 1 | 65536 | 512 | TTFT_SLO_EXCEEDED | None | 16 | 11284.129499543387 | 1 | +| vllm | 8 | 1 | 65536 | 1024 | TTFT_SLO_EXCEEDED | None | 16 | 10645.267280643417 | 1 | +| vllm | 8 | 1 | 131072 | 128 | TTFT_GROUP_SKIP_THRESHOLD | None | 1 | 10573.549484508274 | 1 | + +`Saturation C` is the first point in the final low-gain streak. `Best TPS C` is the tested point with the highest observed Total TPS. diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c112_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c112_i1024_o1024_a1.json new file mode 100644 index 0000000..f5b52f7 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c112_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 560, + "failed": 0, + "duration_s": 143.62631484199665, + "request_tps": 3.899006951588615, + "input_tps": 3992.5831184267417, + "output_tps": 3992.5831184267417, + "total_tps": 7985.1662368534835, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1571.4164454911952, + "ttft_p95_ms": 4123.8772861033785, + "ttft_p99_ms": 4936.747990879667, + "tpot_p50_ms": 25.948950879762915, + "tpot_p95_ms": 28.919239563878318, + "tpot_p99_ms": 29.696389907960754, + "e2e_p50_ms": 28112.99813849473, + "e2e_p95_ms": 31451.82638790211, + "e2e_p99_ms": 31470.12428371294, + "itl_p50_ms": 25.19894449505955, + "itl_p95_ms": 28.669426793931052, + "itl_p99_ms": 31.52743758109864, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o1024_a1.json new file mode 100644 index 0000000..2a947ff --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 70.41322445700644, + "request_tps": 1.1361502134992716, + "input_tps": 1163.417818623254, + "output_tps": 1163.417818623254, + "total_tps": 2326.835637246508, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 421.1962485060212, + "ttft_p95_ms": 677.2944091026147, + "ttft_p99_ms": 691.2584483872342, + "tpot_p50_ms": 12.164721241445186, + "tpot_p95_ms": 15.51125149233277, + "tpot_p99_ms": 15.721278570422475, + "e2e_p50_ms": 12909.268272502231, + "e2e_p95_ms": 16315.028272505877, + "e2e_p99_ms": 16321.569995415775, + "itl_p50_ms": 12.13237400224898, + "itl_p95_ms": 15.468433196656404, + "itl_p99_ms": 16.074650280061178, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o128_a1.json new file mode 100644 index 0000000..6358aeb --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 52.56799242499983, + "request_tps": 1.5218386000594972, + "input_tps": 1558.3627264609252, + "output_tps": 194.79534080761564, + "total_tps": 1753.158067268541, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 937.427844000922, + "ttft_p95_ms": 1428.1466051499592, + "ttft_p99_ms": 18393.34938969926, + "tpot_p50_ms": 19.7043993346339, + "tpot_p95_ms": 181.49350103310894, + "tpot_p99_ms": 182.55337019226835, + "e2e_p50_ms": 3306.429382006172, + "e2e_p95_ms": 23952.145700105757, + "e2e_p99_ms": 23955.10612057842, + "itl_p50_ms": 14.576606998161878, + "itl_p95_ms": 15.284183846233645, + "itl_p99_ms": 20.36303154163767, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o2048_a1.json new file mode 100644 index 0000000..6b3abaf --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 137.77100664599857, + "request_tps": 0.5806736986800084, + "input_tps": 594.6098674483286, + "output_tps": 1189.2197348966572, + "total_tps": 1783.8296023449857, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 509.03518099948997, + "ttft_p95_ms": 817.3343080510676, + "ttft_p99_ms": 821.0908090611338, + "tpot_p50_ms": 12.053855090129739, + "tpot_p95_ms": 15.30965305794184, + "tpot_p99_ms": 15.332859359112767, + "e2e_p50_ms": 25319.949440003256, + "e2e_p95_ms": 31725.24674650267, + "e2e_p99_ms": 31736.754303508642, + "itl_p50_ms": 12.262283504242077, + "itl_p95_ms": 15.428829462325668, + "itl_p99_ms": 15.964222954062278, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o256_a1.json new file mode 100644 index 0000000..4b581d7 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 48.607205292995786, + "request_tps": 1.6458465266162476, + "input_tps": 1685.3468432550376, + "output_tps": 421.3367108137594, + "total_tps": 2106.683554068797, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 782.0743444972322, + "ttft_p95_ms": 12196.891359947767, + "ttft_p99_ms": 12266.554353885149, + "tpot_p50_ms": 15.961954474515354, + "tpot_p95_ms": 58.48537955195449, + "tpot_p99_ms": 60.24822486883116, + "e2e_p50_ms": 8633.398842990573, + "e2e_p95_ms": 16283.070811999643, + "e2e_p99_ms": 16288.124511987555, + "itl_p50_ms": 14.68896050209878, + "itl_p95_ms": 15.627986006438732, + "itl_p99_ms": 16.497610333171902, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o4096_a1.json new file mode 100644 index 0000000..4a790b0 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 271.16081980899617, + "request_tps": 0.29502787333491415, + "input_tps": 302.1085422949521, + "output_tps": 1208.4341691798084, + "total_tps": 1510.5427114747604, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 627.1758095026598, + "ttft_p95_ms": 960.0995247899846, + "ttft_p99_ms": 971.8106476443063, + "tpot_p50_ms": 11.913466125030766, + "tpot_p95_ms": 15.119212534556272, + "tpot_p99_ms": 15.134965707044527, + "e2e_p50_ms": 49374.50365599943, + "e2e_p95_ms": 62329.371708749386, + "e2e_p99_ms": 62340.975158869114, + "itl_p50_ms": 12.052630001562648, + "itl_p95_ms": 15.1747610012535, + "itl_p99_ms": 15.672232958022514, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o512_a1.json new file mode 100644 index 0000000..e41faec --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 37.51114835499902, + "request_tps": 2.132699304294655, + "input_tps": 2183.8840875977266, + "output_tps": 1091.9420437988633, + "total_tps": 3275.82613139659, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 565.4102255066391, + "ttft_p95_ms": 804.2641735068173, + "ttft_p99_ms": 840.2388689920183, + "tpot_p50_ms": 12.695262536184165, + "tpot_p95_ms": 16.153524499127695, + "tpot_p99_ms": 16.279702295489752, + "e2e_p50_ms": 6936.251829996763, + "e2e_p95_ms": 8593.082947545918, + "e2e_p99_ms": 8608.418859792146, + "itl_p50_ms": 12.14090500434395, + "itl_p95_ms": 15.356732595682843, + "itl_p99_ms": 15.861063279880907, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i131072_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i131072_o128_a1.json new file mode 100644 index 0000000..e9fd589 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i131072_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 782.5810133770137, + "request_tps": 0.10222583813371851, + "input_tps": 13398.945055862752, + "output_tps": 13.084907281115969, + "total_tps": 13412.029963143868, + "mean_input_tokens": 131072.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 58004.37130450155, + "ttft_p95_ms": 110342.86055744666, + "ttft_p99_ms": 145009.39971607734, + "tpot_p50_ms": 726.4293738111247, + "tpot_p95_ms": 1148.8756013172838, + "tpot_p99_ms": 1185.2487841417542, + "e2e_p50_ms": 145932.8185465056, + "e2e_p95_ms": 251712.73757864692, + "e2e_p99_ms": 293697.5071116265, + "itl_p50_ms": 1063.0883960038773, + "itl_p95_ms": 1341.7487418533708, + "itl_p99_ms": 1398.4825665256358, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o1024_a1.json new file mode 100644 index 0000000..bb66243 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 81.79191887499474, + "request_tps": 0.9780917369387875, + "input_tps": 16025.055018005094, + "output_tps": 1001.5659386253184, + "total_tps": 17026.620956630413, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 999.8386475053849, + "ttft_p95_ms": 3296.2886180946934, + "ttft_p99_ms": 4053.4733163932106, + "tpot_p50_ms": 15.505134818183233, + "tpot_p95_ms": 18.54402336778967, + "tpot_p99_ms": 18.632150808406323, + "e2e_p50_ms": 16877.44059450779, + "e2e_p95_ms": 19530.405244355643, + "e2e_p99_ms": 19544.808635528025, + "itl_p50_ms": 15.109817002667114, + "itl_p95_ms": 16.077759441395756, + "itl_p99_ms": 16.771573076403005, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o128_a1.json new file mode 100644 index 0000000..b6d315e --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 61.62385746999644, + "request_tps": 1.2981985108438008, + "input_tps": 21269.684401664832, + "output_tps": 166.1694093880065, + "total_tps": 21435.85381105284, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 6200.594059002469, + "ttft_p95_ms": 9982.714110245432, + "ttft_p99_ms": 12083.678980552644, + "tpot_p50_ms": 46.56020286221444, + "tpot_p95_ms": 83.96291618064566, + "tpot_p99_ms": 94.82308160958443, + "e2e_p50_ms": 11807.360419501492, + "e2e_p95_ms": 14518.52319239406, + "e2e_p99_ms": 15406.82603497858, + "itl_p50_ms": 14.998138998635113, + "itl_p95_ms": 19.290680799167607, + "itl_p99_ms": 913.8215772807598, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o2048_a1.json new file mode 100644 index 0000000..0c26de3 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 136.13383251799678, + "request_tps": 0.5876570028205448, + "input_tps": 9628.172334211806, + "output_tps": 1203.5215417764757, + "total_tps": 10831.693875988281, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 952.042529002938, + "ttft_p95_ms": 1976.2296090004384, + "ttft_p99_ms": 2348.7864647468086, + "tpot_p50_ms": 12.17877617391463, + "tpot_p95_ms": 15.498852642595702, + "tpot_p99_ms": 15.918200653907457, + "e2e_p50_ms": 25627.524990501115, + "e2e_p95_ms": 32975.65399566083, + "e2e_p99_ms": 32987.294595711865, + "itl_p50_ms": 11.984888005827088, + "itl_p95_ms": 15.32018614525441, + "itl_p99_ms": 15.952927653561346, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o256_a1.json new file mode 100644 index 0000000..c5b79c8 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 48.023830674996134, + "request_tps": 1.665839623277958, + "input_tps": 27293.116387786064, + "output_tps": 426.45494355915724, + "total_tps": 27719.57133134522, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 4218.072452502383, + "ttft_p95_ms": 7202.70946295495, + "ttft_p99_ms": 9814.834247549588, + "tpot_p50_ms": 20.867704264716902, + "tpot_p95_ms": 43.5013626262719, + "tpot_p99_ms": 48.00088939475241, + "e2e_p50_ms": 10424.10009400919, + "e2e_p95_ms": 13663.989264789414, + "e2e_p99_ms": 16127.52558444379, + "itl_p50_ms": 14.975312995375134, + "itl_p95_ms": 15.900831998442298, + "itl_p99_ms": 664.3842326034808, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o512_a1.json new file mode 100644 index 0000000..72edd0c --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i16384_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 56.001584703000844, + "request_tps": 1.4285310036184244, + "input_tps": 23405.051963284266, + "output_tps": 731.4078738526333, + "total_tps": 24136.4598371369, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 2150.0688629894285, + "ttft_p95_ms": 4893.693922802049, + "ttft_p99_ms": 5737.629971327547, + "tpot_p50_ms": 16.641745551845485, + "tpot_p95_ms": 22.60100227660779, + "tpot_p99_ms": 25.976594107668582, + "e2e_p50_ms": 10941.78693700087, + "e2e_p95_ms": 14099.559878905711, + "e2e_p99_ms": 14118.464289340336, + "itl_p50_ms": 15.120271986234002, + "itl_p95_ms": 16.094310994958505, + "itl_p99_ms": 19.090893008979037, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o1024_a1.json new file mode 100644 index 0000000..80ea27b --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 74.54846192999685, + "request_tps": 1.0731274385663692, + "input_tps": 4395.529988367848, + "output_tps": 1098.882497091962, + "total_tps": 5494.41248545981, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 664.4156365000526, + "ttft_p95_ms": 974.0240385996003, + "ttft_p99_ms": 982.1483760114643, + "tpot_p50_ms": 14.938762804988556, + "tpot_p95_ms": 15.543498240374383, + "tpot_p99_ms": 15.80483324938253, + "e2e_p50_ms": 16048.860944996704, + "e2e_p95_ms": 16527.93401440722, + "e2e_p99_ms": 16534.28529751647, + "itl_p50_ms": 14.674316000309773, + "itl_p95_ms": 15.759427801822312, + "itl_p99_ms": 16.294228279730305, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o128_a1.json new file mode 100644 index 0000000..aea7cd4 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 63.706278053999995, + "request_tps": 1.2557632064486453, + "input_tps": 5143.606093613651, + "output_tps": 160.7376904254266, + "total_tps": 5304.343784039078, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 2722.8040709960624, + "ttft_p95_ms": 26840.74661414561, + "ttft_p99_ms": 27275.76444137885, + "tpot_p50_ms": 20.65726973231667, + "tpot_p95_ms": 151.34337404136696, + "tpot_p99_ms": 207.36058258637357, + "e2e_p50_ms": 5354.688991996227, + "e2e_p95_ms": 29170.501739544852, + "e2e_p99_ms": 29177.230598040624, + "itl_p50_ms": 14.814941496297251, + "itl_p95_ms": 15.711286505393218, + "itl_p99_ms": 182.99884561245784, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o2048_a1.json new file mode 100644 index 0000000..565a122 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 138.75295940799697, + "request_tps": 0.5765642789986448, + "input_tps": 2361.607286778449, + "output_tps": 1180.8036433892246, + "total_tps": 3542.4109301676735, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 632.4307624963694, + "ttft_p95_ms": 959.6396220949828, + "ttft_p99_ms": 969.5783248801308, + "tpot_p50_ms": 12.087159364439152, + "tpot_p95_ms": 15.38081124682696, + "tpot_p99_ms": 15.53437136937437, + "e2e_p50_ms": 25168.70826199738, + "e2e_p95_ms": 32143.357299292988, + "e2e_p99_ms": 32148.356982687113, + "itl_p50_ms": 12.129640497732908, + "itl_p95_ms": 15.666113744373433, + "itl_p99_ms": 16.459487642714517, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o256_a1.json new file mode 100644 index 0000000..3cbe33f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 25.900737181000295, + "request_tps": 3.088715175979033, + "input_tps": 12651.377360810118, + "output_tps": 790.7110850506324, + "total_tps": 13442.08844586075, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 1070.6434219973744, + "ttft_p95_ms": 2601.758133142721, + "ttft_p99_ms": 2731.048192037851, + "tpot_p50_ms": 14.66723954899684, + "tpot_p95_ms": 20.23464126315529, + "tpot_p99_ms": 21.929446337492916, + "e2e_p50_ms": 5262.183128004835, + "e2e_p95_ms": 6524.885636250838, + "e2e_p99_ms": 6530.703649207862, + "itl_p50_ms": 12.11705449532019, + "itl_p95_ms": 15.253629755170547, + "itl_p99_ms": 16.63169343562916, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o4096_a1.json new file mode 100644 index 0000000..03f1a35 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 274.1387693459983, + "request_tps": 0.2918230069787383, + "input_tps": 1195.307036584912, + "output_tps": 1195.307036584912, + "total_tps": 2390.614073169824, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1043.4461850018124, + "ttft_p95_ms": 1874.445231048594, + "ttft_p99_ms": 1897.9588461619276, + "tpot_p50_ms": 11.862804140170747, + "tpot_p95_ms": 15.333563006644244, + "tpot_p99_ms": 15.394116480524199, + "e2e_p50_ms": 49251.109377997636, + "e2e_p95_ms": 63309.41746450044, + "e2e_p99_ms": 63314.34459721451, + "itl_p50_ms": 11.962887001573108, + "itl_p95_ms": 15.223909802443814, + "itl_p99_ms": 15.640087766805664, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o512_a1.json new file mode 100644 index 0000000..fc97182 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i4096_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 37.81364577100612, + "request_tps": 2.1156383725723837, + "input_tps": 8665.654774056484, + "output_tps": 1083.2068467570605, + "total_tps": 9748.861620813545, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 626.4680629974464, + "ttft_p95_ms": 1079.5922750941827, + "ttft_p99_ms": 1085.4407174367225, + "tpot_p50_ms": 12.883891402167945, + "tpot_p95_ms": 15.779685464180407, + "tpot_p99_ms": 16.333569430081702, + "e2e_p50_ms": 7039.921916504682, + "e2e_p95_ms": 8655.18560165001, + "e2e_p99_ms": 8659.915602506953, + "itl_p50_ms": 12.184623003122397, + "itl_p95_ms": 15.375507500721142, + "itl_p99_ms": 16.372877460671585, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o1024_a1.json new file mode 100644 index 0000000..56b3dfc --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 390.9239437669894, + "request_tps": 0.2046433872254294, + "input_tps": 13411.509025205742, + "output_tps": 209.5548285188397, + "total_tps": 13621.063853724581, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 34109.15530550119, + "ttft_p95_ms": 49910.73756714177, + "ttft_p99_ms": 58476.69062248939, + "tpot_p50_ms": 42.09774688367832, + "tpot_p95_ms": 69.21228409907829, + "tpot_p99_ms": 69.88106847383611, + "e2e_p50_ms": 79284.3169455009, + "e2e_p95_ms": 90811.3658934919, + "e2e_p99_ms": 100775.45445886119, + "itl_p50_ms": 12.344260001555085, + "itl_p95_ms": 13.471613003639504, + "itl_p99_ms": 1075.7671469193883, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o128_a1.json new file mode 100644 index 0000000..3393c35 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 332.67608038699836, + "request_tps": 0.24047415704470515, + "input_tps": 15759.714356081797, + "output_tps": 30.78069210172226, + "total_tps": 15790.495048183519, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 31602.030366491817, + "ttft_p95_ms": 50072.25141515082, + "ttft_p99_ms": 58527.80881931147, + "tpot_p50_ms": 269.0473995314715, + "tpot_p95_ms": 453.42705731063467, + "tpot_p99_ms": 478.3763391067141, + "e2e_p50_ms": 67748.1715354952, + "e2e_p95_ms": 83297.61391548891, + "e2e_p99_ms": 92202.8402509741, + "itl_p50_ms": 12.720453007204924, + "itl_p95_ms": 1088.2944416043756, + "itl_p99_ms": 1121.3325642509153, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o256_a1.json new file mode 100644 index 0000000..513eace --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 340.1405706650112, + "request_tps": 0.23519687711345766, + "input_tps": 15413.862538507561, + "output_tps": 60.21040054104516, + "total_tps": 15474.072939048607, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 33961.24007800245, + "ttft_p95_ms": 49937.02013984801, + "ttft_p99_ms": 62166.142975594805, + "tpot_p50_ms": 139.35726555489464, + "tpot_p95_ms": 236.3478919172656, + "tpot_p99_ms": 244.71531864295898, + "e2e_p50_ms": 69747.46793750091, + "e2e_p95_ms": 78453.58004349907, + "e2e_p99_ms": 96399.40396136702, + "itl_p50_ms": 12.44720199611038, + "itl_p95_ms": 1048.4594318972086, + "itl_p99_ms": 1099.1333858296275, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o512_a1.json new file mode 100644 index 0000000..7cfb3f2 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c16_i65536_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 365.2319012020016, + "request_tps": 0.21903891674499099, + "input_tps": 14354.93444779973, + "output_tps": 112.14792537343538, + "total_tps": 14467.082373173163, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 33553.05798899644, + "ttft_p95_ms": 49813.80844564191, + "ttft_p99_ms": 59237.044557964655, + "tpot_p50_ms": 79.3469260440206, + "tpot_p95_ms": 126.30341421105851, + "tpot_p99_ms": 137.98316977467067, + "e2e_p50_ms": 73173.0037334928, + "e2e_p95_ms": 90507.30776954733, + "e2e_p99_ms": 102572.02681419696, + "itl_p50_ms": 12.42508299765177, + "itl_p95_ms": 974.080784598482, + "itl_p99_ms": 1092.6132275909183, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i131072_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i131072_o128_a1.json new file mode 100644 index 0000000..2901a25 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i131072_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 82.18055039300816, + "request_tps": 0.06084164654639981, + "input_tps": 7974.636296129715, + "output_tps": 7.787730757939175, + "total_tps": 7982.424026887654, + "mean_input_tokens": 131072.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 15437.700795999262, + "ttft_p95_ms": 15571.00856200559, + "ttft_p99_ms": 15593.639437204693, + "tpot_p50_ms": 7.764350944867827, + "tpot_p95_ms": 7.770626307061807, + "tpot_p99_ms": 7.771325803097836, + "e2e_p50_ms": 16424.126157013234, + "e2e_p95_ms": 16554.536877613282, + "e2e_p99_ms": 16576.65890193428, + "itl_p50_ms": 8.529996004654095, + "itl_p95_ms": 8.682068198686466, + "itl_p99_ms": 9.645803358289447, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o1024_a1.json new file mode 100644 index 0000000..89d3e32 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 76.87229951999325, + "request_tps": 0.06504293524743045, + "input_tps": 4262.653804375602, + "output_tps": 66.60396569336878, + "total_tps": 4329.257770068971, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 6793.178814987186, + "ttft_p95_ms": 6830.732189002447, + "ttft_p99_ms": 6836.927511403337, + "tpot_p50_ms": 8.389067413488618, + "tpot_p95_ms": 8.404996331775118, + "tpot_p99_ms": 8.405754934785953, + "e2e_p50_ms": 15372.661996996612, + "e2e_p95_ms": 15412.706287001492, + "e2e_p99_ms": 15418.935102202231, + "itl_p50_ms": 8.452879003016278, + "itl_p95_ms": 8.561783994082361, + "itl_p99_ms": 8.750601002248002, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o128_a1.json new file mode 100644 index 0000000..e5c1a31 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 38.88842260101228, + "request_tps": 0.12857297019472957, + "input_tps": 8426.158174681797, + "output_tps": 16.457340184925386, + "total_tps": 8442.615514866722, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 6736.3390589889605, + "ttft_p95_ms": 6827.516746404581, + "ttft_p99_ms": 6834.052311684936, + "tpot_p50_ms": 7.981791047220794, + "tpot_p95_ms": 8.017796259758422, + "tpot_p99_ms": 8.0200000455801, + "e2e_p50_ms": 7747.252895991551, + "e2e_p95_ms": 7842.1887119999155, + "e2e_p99_ms": 7847.936675201636, + "itl_p50_ms": 8.454770999378525, + "itl_p95_ms": 8.66287379467394, + "itl_p99_ms": 10.578372052987115, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o256_a1.json new file mode 100644 index 0000000..ec541c8 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 44.46872818299744, + "request_tps": 0.11243856535370274, + "input_tps": 7368.773819020263, + "output_tps": 28.7842727305479, + "total_tps": 7397.5580917508105, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 6767.2624370025005, + "ttft_p95_ms": 6892.302849600674, + "ttft_p99_ms": 6911.403660320793, + "tpot_p50_ms": 8.220937709819397, + "tpot_p95_ms": 8.2882254510004, + "tpot_p99_ms": 8.295880123954632, + "e2e_p50_ms": 8871.000218990957, + "e2e_p95_ms": 9004.320606408874, + "e2e_p99_ms": 9026.557145289844, + "itl_p50_ms": 8.483409998007119, + "itl_p95_ms": 8.675873003085144, + "itl_p99_ms": 9.017845205380567, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o512_a1.json new file mode 100644 index 0000000..337d8da --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c1_i65536_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 55.36560429999372, + "request_tps": 0.09030877677967596, + "input_tps": 5918.4759950328435, + "output_tps": 46.23809371119409, + "total_tps": 5964.714088744037, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 6785.467125999276, + "ttft_p95_ms": 6867.230469404603, + "ttft_p99_ms": 6880.844480285305, + "tpot_p50_ms": 8.349586817987925, + "tpot_p95_ms": 8.493367793736068, + "tpot_p99_ms": 8.520835669118448, + "e2e_p50_ms": 11042.787049998879, + "e2e_p95_ms": 11206.68308060267, + "e2e_p99_ms": 11234.859840924619, + "itl_p50_ms": 8.474283000396099, + "itl_p95_ms": 8.707773692003684, + "itl_p99_ms": 9.049925783474462, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o1024_a1.json new file mode 100644 index 0000000..e47b46d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 85.3312213080062, + "request_tps": 1.875046407955115, + "input_tps": 1920.0475217460378, + "output_tps": 1920.0475217460378, + "total_tps": 3840.0950434920755, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 806.487787005608, + "ttft_p95_ms": 900.6144404585939, + "ttft_p99_ms": 904.4747244215978, + "tpot_p50_ms": 15.241720033234294, + "tpot_p95_ms": 17.703688128979007, + "tpot_p99_ms": 17.972658989592805, + "e2e_p50_ms": 16181.125543495, + "e2e_p95_ms": 18731.30079265975, + "e2e_p99_ms": 18746.509167428594, + "itl_p50_ms": 15.200200999970548, + "itl_p95_ms": 17.96690479677636, + "itl_p99_ms": 18.968004560447298, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o128_a1.json new file mode 100644 index 0000000..b352162 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 95.59839020299842, + "request_tps": 1.6736683500658116, + "input_tps": 1713.836390467391, + "output_tps": 214.22954880842389, + "total_tps": 1928.0659392758148, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 4119.889060501009, + "ttft_p95_ms": 26025.381604798167, + "ttft_p99_ms": 26478.019099681696, + "tpot_p50_ms": 21.413118271642976, + "tpot_p95_ms": 376.97693987998275, + "tpot_p99_ms": 387.07957727625, + "e2e_p50_ms": 7732.499119003478, + "e2e_p95_ms": 51129.2892126934, + "e2e_p99_ms": 51135.8539534836, + "itl_p50_ms": 16.923131996009033, + "itl_p95_ms": 18.388393107306907, + "itl_p99_ms": 244.194259904325, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o2048_a1.json new file mode 100644 index 0000000..c82ff70 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 162.71412774299097, + "request_tps": 0.9833196552712499, + "input_tps": 1006.9193269977599, + "output_tps": 2013.8386539955197, + "total_tps": 3020.75798099328, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 871.0656835028203, + "ttft_p95_ms": 1184.4237971468829, + "ttft_p99_ms": 1192.6747960213106, + "tpot_p50_ms": 15.119994677330888, + "tpot_p95_ms": 17.311258725843935, + "tpot_p99_ms": 17.470737866078203, + "e2e_p50_ms": 31956.353274501453, + "e2e_p95_ms": 36126.69280984701, + "e2e_p99_ms": 36139.65432776953, + "itl_p50_ms": 14.982249005697668, + "itl_p95_ms": 17.40834979864303, + "itl_p99_ms": 18.12914603331592, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o4096_a1.json new file mode 100644 index 0000000..f475c20 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 318.9912624319986, + "request_tps": 0.50158113667489, + "input_tps": 513.6190839550874, + "output_tps": 2054.4763358203495, + "total_tps": 2568.0954197754368, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 867.0998999950825, + "ttft_p95_ms": 1568.6099555969122, + "ttft_p99_ms": 1761.4919271026156, + "tpot_p50_ms": 14.940362645543985, + "tpot_p95_ms": 17.25645533934084, + "tpot_p99_ms": 17.34749941946349, + "e2e_p50_ms": 62241.984514999785, + "e2e_p95_ms": 71418.52431923835, + "e2e_p99_ms": 71433.78462005581, + "itl_p50_ms": 14.883934498357121, + "itl_p95_ms": 17.468459051451646, + "itl_p99_ms": 18.707020131550955, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o512_a1.json new file mode 100644 index 0000000..73d27d4 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 44.825771888004965, + "request_tps": 3.569375233509694, + "input_tps": 3655.040239113927, + "output_tps": 1827.5201195569634, + "total_tps": 5482.56035867089, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 912.5870395073434, + "ttft_p95_ms": 1430.6775802040647, + "ttft_p99_ms": 1454.0497396518183, + "tpot_p50_ms": 15.280834627210176, + "tpot_p95_ms": 18.225302902624872, + "tpot_p99_ms": 18.700063700290844, + "e2e_p50_ms": 8596.120701506152, + "e2e_p95_ms": 10176.577183793415, + "e2e_p99_ms": 10183.192389139585, + "itl_p50_ms": 14.963374997023493, + "itl_p95_ms": 17.358059901744127, + "itl_p99_ms": 18.431273359310598, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o1024_a1.json new file mode 100644 index 0000000..c24929f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 165.0628843260056, + "request_tps": 0.969327542368603, + "input_tps": 15881.462454167191, + "output_tps": 992.5914033854494, + "total_tps": 16874.05385755264, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 4267.802063499403, + "ttft_p95_ms": 19862.928376002674, + "ttft_p99_ms": 21657.044113324228, + "tpot_p50_ms": 25.376833950635163, + "tpot_p95_ms": 37.8741763598776, + "tpot_p99_ms": 40.72076095410008, + "e2e_p50_ms": 35063.07619450672, + "e2e_p95_ms": 51894.37677649694, + "e2e_p99_ms": 53688.66714284828, + "itl_p50_ms": 17.057024502719287, + "itl_p95_ms": 18.18839350307826, + "itl_p99_ms": 674.1142390163444, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o2048_a1.json new file mode 100644 index 0000000..b985e62 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i16384_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 203.0028582619998, + "request_tps": 0.7881662424353681, + "input_tps": 12913.315716061072, + "output_tps": 1614.164464507634, + "total_tps": 14527.480180568706, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 2550.8961580053438, + "ttft_p95_ms": 15365.738819241233, + "ttft_p99_ms": 18318.440213190624, + "tpot_p50_ms": 16.89672226160163, + "tpot_p95_ms": 21.461009534389174, + "tpot_p99_ms": 22.790276562500463, + "e2e_p50_ms": 40052.36150848941, + "e2e_p95_ms": 48792.51678490764, + "e2e_p99_ms": 48817.223941341945, + "itl_p50_ms": 14.973600002122112, + "itl_p95_ms": 17.530049994093133, + "itl_p99_ms": 18.100957148271835, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o1024_a1.json new file mode 100644 index 0000000..e3ac3a2 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 86.79001112899277, + "request_tps": 1.843530124246648, + "input_tps": 7551.09938891427, + "output_tps": 1887.7748472285675, + "total_tps": 9438.874236142838, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1054.5200394990388, + "ttft_p95_ms": 1617.5321672504651, + "ttft_p99_ms": 1622.538609701296, + "tpot_p50_ms": 15.406173739000165, + "tpot_p95_ms": 17.669181664998092, + "tpot_p99_ms": 18.132834118727168, + "e2e_p50_ms": 16632.105807497283, + "e2e_p95_ms": 18881.63431045614, + "e2e_p99_ms": 18892.737503780663, + "itl_p50_ms": 15.119137002329808, + "itl_p95_ms": 17.811838999477914, + "itl_p99_ms": 18.673117576690856, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o2048_a1.json new file mode 100644 index 0000000..20ce709 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 171.35218937399623, + "request_tps": 0.9337493765590661, + "input_tps": 3824.6374463859347, + "output_tps": 1912.3187231929674, + "total_tps": 5736.956169578902, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 897.8936719940975, + "ttft_p95_ms": 1402.46574630728, + "ttft_p99_ms": 1414.9006537273815, + "tpot_p50_ms": 17.125081185391775, + "tpot_p95_ms": 17.535651223030666, + "tpot_p99_ms": 17.64257730764133, + "e2e_p50_ms": 35692.707454494666, + "e2e_p95_ms": 36474.77745984506, + "e2e_p99_ms": 36482.97110349682, + "itl_p50_ms": 16.84335399477277, + "itl_p95_ms": 17.664605999016203, + "itl_p99_ms": 18.40053380583413, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o256_a1.json new file mode 100644 index 0000000..424ea56 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 39.068993989989394, + "request_tps": 4.0953191689808195, + "input_tps": 16774.427316145437, + "output_tps": 1048.4017072590898, + "total_tps": 17822.829023404527, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 2343.7372504995437, + "ttft_p95_ms": 5162.197779646522, + "ttft_p99_ms": 5491.549329505941, + "tpot_p50_ms": 19.213523033335253, + "tpot_p95_ms": 32.66720932803069, + "tpot_p99_ms": 34.66762047272469, + "e2e_p50_ms": 8660.10677999293, + "e2e_p95_ms": 9275.378481158987, + "e2e_p99_ms": 9288.702330155647, + "itl_p50_ms": 14.975201003835537, + "itl_p95_ms": 17.6850339921657, + "itl_p99_ms": 28.16523999499614, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o4096_a1.json new file mode 100644 index 0000000..7c8679a --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 361.17721821699524, + "request_tps": 0.442995825677665, + "input_tps": 1814.5109019757158, + "output_tps": 1814.5109019757158, + "total_tps": 3629.0218039514316, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1892.501820497273, + "ttft_p95_ms": 4395.493246797548, + "ttft_p99_ms": 5224.942043306945, + "tpot_p50_ms": 17.487011898045463, + "tpot_p95_ms": 18.118140513893067, + "tpot_p99_ms": 18.395078147048498, + "e2e_p50_ms": 74226.02639300021, + "e2e_p95_ms": 75991.18075425722, + "e2e_p99_ms": 76000.55798450441, + "itl_p50_ms": 17.170556006021798, + "itl_p95_ms": 18.040875489532482, + "itl_p99_ms": 18.916477599123027, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o512_a1.json new file mode 100644 index 0000000..e4c3a94 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c32_i4096_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 56.462142722986755, + "request_tps": 2.83375713856607, + "input_tps": 11607.069239566623, + "output_tps": 1450.8836549458279, + "total_tps": 13057.952894512451, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1453.6274200072512, + "ttft_p95_ms": 3663.0988704906445, + "ttft_p99_ms": 4404.091445246595, + "tpot_p50_ms": 18.000749555776817, + "tpot_p95_ms": 23.179044860871425, + "tpot_p99_ms": 23.890643953811313, + "e2e_p50_ms": 11132.362890006334, + "e2e_p95_ms": 13523.678070852475, + "e2e_p99_ms": 13540.85523293179, + "itl_p50_ms": 17.186335011501797, + "itl_p95_ms": 18.29535180295352, + "itl_p99_ms": 19.591979318647628, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o1024_a1.json new file mode 100644 index 0000000..5ff9e9f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 97.32574620700325, + "request_tps": 2.46594564494313, + "input_tps": 2525.128340421765, + "output_tps": 2525.128340421765, + "total_tps": 5050.25668084353, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 971.1724649969256, + "ttft_p95_ms": 1047.563062050176, + "ttft_p99_ms": 1059.1346113267355, + "tpot_p50_ms": 17.463300068922933, + "tpot_p95_ms": 20.0403643846501, + "tpot_p99_ms": 20.23941422052442, + "e2e_p50_ms": 18865.876165502414, + "e2e_p95_ms": 20948.537042703538, + "e2e_p99_ms": 20966.204005898326, + "itl_p50_ms": 17.521906993351877, + "itl_p95_ms": 19.884526249370538, + "itl_p99_ms": 20.897341451927787, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o2048_a1.json new file mode 100644 index 0000000..d3803fc --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 205.12268571999448, + "request_tps": 1.170031482171676, + "input_tps": 1198.1122377437962, + "output_tps": 2396.2244754875924, + "total_tps": 3594.3367132313883, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1103.477151504194, + "ttft_p95_ms": 1655.3460868577529, + "ttft_p99_ms": 2063.1610584430746, + "tpot_p50_ms": 19.476409309720992, + "tpot_p95_ms": 19.98451202838762, + "tpot_p99_ms": 20.224875959688827, + "e2e_p50_ms": 41019.6206035107, + "e2e_p95_ms": 41983.93201486106, + "e2e_p99_ms": 42004.45509949772, + "itl_p50_ms": 19.33959749294445, + "itl_p95_ms": 21.063311200850876, + "itl_p99_ms": 22.709367697680136, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o4096_a1.json new file mode 100644 index 0000000..5c887f3 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 387.6589750939893, + "request_tps": 0.6191008474441, + "input_tps": 633.9592677827584, + "output_tps": 2535.8370711310336, + "total_tps": 3169.7963389137917, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1218.414176008082, + "ttft_p95_ms": 2534.173024399205, + "ttft_p99_ms": 2783.003787570633, + "tpot_p50_ms": 19.216241195482507, + "tpot_p95_ms": 19.885012049902812, + "tpot_p99_ms": 19.88576329422029, + "e2e_p50_ms": 79752.23233650468, + "e2e_p95_ms": 82442.38089720166, + "e2e_p99_ms": 82460.91719069882, + "itl_p50_ms": 18.868017999920994, + "itl_p95_ms": 20.22105274882051, + "itl_p99_ms": 21.84910236479482, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o512_a1.json new file mode 100644 index 0000000..78a9d8b --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 57.3580158239929, + "request_tps": 4.184245158278432, + "input_tps": 4284.667042077114, + "output_tps": 2142.333521038557, + "total_tps": 6427.000563115672, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1114.938983497268, + "ttft_p95_ms": 2623.5857654435677, + "ttft_p99_ms": 2657.5968347305025, + "tpot_p50_ms": 19.735206197671033, + "tpot_p95_ms": 22.902151353534165, + "tpot_p99_ms": 23.980957457851613, + "e2e_p50_ms": 11155.085417994997, + "e2e_p95_ms": 12560.576499555464, + "e2e_p99_ms": 12733.288235523505, + "itl_p50_ms": 19.239879009546712, + "itl_p95_ms": 20.936376793542877, + "itl_p99_ms": 22.45631740195677, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o1024_a1.json new file mode 100644 index 0000000..5ea33ae --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 107.01144343199849, + "request_tps": 2.242750796577288, + "input_tps": 9186.307262780572, + "output_tps": 2296.576815695143, + "total_tps": 11482.884078475714, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1429.1901315009454, + "ttft_p95_ms": 4655.2081622539845, + "ttft_p99_ms": 5421.287226663553, + "tpot_p50_ms": 17.961494444764135, + "tpot_p95_ms": 23.42196408294526, + "tpot_p99_ms": 23.725152669697756, + "e2e_p50_ms": 19863.534916497883, + "e2e_p95_ms": 25332.92310884499, + "e2e_p99_ms": 25341.139651168924, + "itl_p50_ms": 17.542485002195463, + "itl_p95_ms": 19.82468349451665, + "itl_p99_ms": 22.084278145339344, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o2048_a1.json new file mode 100644 index 0000000..b810c05 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 195.2795769630029, + "request_tps": 1.2290071687602524, + "input_tps": 5034.013363241994, + "output_tps": 2517.006681620997, + "total_tps": 7551.02004486299, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1402.5059129999136, + "ttft_p95_ms": 3014.964500940548, + "ttft_p99_ms": 3645.231739671726, + "tpot_p50_ms": 17.624665780901058, + "tpot_p95_ms": 20.07420706065267, + "tpot_p99_ms": 20.200438402067306, + "e2e_p50_ms": 38203.180718497606, + "e2e_p95_ms": 42234.80308325525, + "e2e_p99_ms": 42259.54538603983, + "itl_p50_ms": 17.31949599343352, + "itl_p95_ms": 20.04093239957001, + "itl_p99_ms": 21.313599089044146, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o512_a1.json new file mode 100644 index 0000000..6e6a387 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c48_i4096_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 78.62980660999892, + "request_tps": 3.052277632964196, + "input_tps": 12502.129184621346, + "output_tps": 1562.7661480776683, + "total_tps": 14064.895332699014, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 2010.3960574997473, + "ttft_p95_ms": 8091.9554406515035, + "ttft_p99_ms": 9461.760722859906, + "tpot_p50_ms": 23.111368014671182, + "tpot_p95_ms": 34.88897108324944, + "tpot_p99_ms": 38.233606207983335, + "e2e_p50_ms": 14650.032247503987, + "e2e_p95_ms": 20735.955423545238, + "e2e_p99_ms": 21034.978971217053, + "itl_p50_ms": 19.645875494461507, + "itl_p95_ms": 22.24900509609142, + "itl_p99_ms": 25.97859771209185, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o1024_a1.json new file mode 100644 index 0000000..d7abd07 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 111.44104069900641, + "request_tps": 2.8714735432549947, + "input_tps": 2940.3889082931146, + "output_tps": 2940.3889082931146, + "total_tps": 5880.777816586229, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1251.2050785007887, + "ttft_p95_ms": 1664.9364830955165, + "ttft_p99_ms": 1687.642185555742, + "tpot_p50_ms": 20.892368739980142, + "tpot_p95_ms": 22.107194719502488, + "tpot_p99_ms": 22.406787545171895, + "e2e_p50_ms": 22625.58904299658, + "e2e_p95_ms": 23358.138840601896, + "e2e_p99_ms": 23385.716578069405, + "itl_p50_ms": 20.738980994792655, + "itl_p95_ms": 21.87385440629441, + "itl_p99_ms": 24.061159000266343, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o2048_a1.json new file mode 100644 index 0000000..a94e6b3 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 227.52200377998815, + "request_tps": 1.4064573741598956, + "input_tps": 1440.212351139733, + "output_tps": 2880.424702279466, + "total_tps": 4320.637053419199, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1391.4943349955138, + "ttft_p95_ms": 2766.4422078472853, + "ttft_p99_ms": 3516.292367083952, + "tpot_p50_ms": 21.495228763068177, + "tpot_p95_ms": 22.537871907642682, + "tpot_p99_ms": 22.736073239234297, + "e2e_p50_ms": 45078.62572550221, + "e2e_p95_ms": 47292.98801284676, + "e2e_p99_ms": 47307.55221481537, + "itl_p50_ms": 21.152492990950122, + "itl_p95_ms": 22.822722894488834, + "itl_p99_ms": 24.930158918141395, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o4096_a1.json new file mode 100644 index 0000000..68692fe --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 429.61134696399677, + "request_tps": 0.7448592833066333, + "input_tps": 762.7359061059925, + "output_tps": 3050.94362442397, + "total_tps": 3813.6795305299624, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1419.7698949938058, + "ttft_p95_ms": 2921.583319955971, + "ttft_p99_ms": 3675.6699006249255, + "tpot_p50_ms": 20.945350810012243, + "tpot_p95_ms": 21.617453843995357, + "tpot_p99_ms": 21.769692606291244, + "e2e_p50_ms": 87146.0334000003, + "e2e_p95_ms": 89824.40359709944, + "e2e_p99_ms": 89989.40303418975, + "itl_p50_ms": 20.697043000836857, + "itl_p95_ms": 22.144093887618506, + "itl_p99_ms": 24.082228924962692, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o512_a1.json new file mode 100644 index 0000000..9664176 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 64.12715022399789, + "request_tps": 4.990086084945788, + "input_tps": 5109.848150984487, + "output_tps": 2554.9240754922434, + "total_tps": 7664.77222647673, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1574.427007995837, + "ttft_p95_ms": 2990.9389156964608, + "ttft_p99_ms": 3418.0432158199255, + "tpot_p50_ms": 21.48998137866181, + "tpot_p95_ms": 26.06068395949076, + "tpot_p99_ms": 26.0992564425009, + "e2e_p50_ms": 13091.199074500764, + "e2e_p95_ms": 14394.905802008725, + "e2e_p99_ms": 14404.70782280463, + "itl_p50_ms": 20.648846999392845, + "itl_p95_ms": 22.798411999247037, + "itl_p99_ms": 25.05893900524825, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i4096_o2048_a1.json new file mode 100644 index 0000000..cbef255 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c64_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 240.90383681698586, + "request_tps": 1.3283308569431518, + "input_tps": 5440.84319003915, + "output_tps": 2720.421595019575, + "total_tps": 8161.264785058724, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1544.8923770018155, + "ttft_p95_ms": 9636.169896001238, + "ttft_p99_ms": 11990.489179552387, + "tpot_p50_ms": 21.333603897412154, + "tpot_p95_ms": 26.558363334025465, + "tpot_p99_ms": 27.72400532863237, + "e2e_p50_ms": 46171.67401249753, + "e2e_p95_ms": 58603.167127201596, + "e2e_p99_ms": 58643.71314777425, + "itl_p50_ms": 20.934874992235564, + "itl_p95_ms": 22.243883497139905, + "itl_p99_ms": 24.41439779868233, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o1024_a1.json new file mode 100644 index 0000000..4bd7e38 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 400, + "failed": 0, + "duration_s": 123.47737352400145, + "request_tps": 3.2394598992846917, + "input_tps": 3317.2069368675243, + "output_tps": 3317.2069368675243, + "total_tps": 6634.413873735049, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1259.8191590004717, + "ttft_p95_ms": 2176.3979704985104, + "ttft_p99_ms": 2561.1882165598217, + "tpot_p50_ms": 23.053262887589092, + "tpot_p95_ms": 23.774626320429117, + "tpot_p99_ms": 24.000006763245814, + "e2e_p50_ms": 24829.307881504064, + "e2e_p95_ms": 24994.919799604395, + "e2e_p99_ms": 25020.736574648326, + "itl_p50_ms": 22.634291992289945, + "itl_p95_ms": 24.973727005999535, + "itl_p99_ms": 27.328714399482124, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o2048_a1.json new file mode 100644 index 0000000..ea05462 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 400, + "failed": 0, + "duration_s": 265.8902801889926, + "request_tps": 1.5043799258689836, + "input_tps": 1540.4850440898392, + "output_tps": 3080.9700881796784, + "total_tps": 4621.455132269518, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1453.8656489967252, + "ttft_p95_ms": 4220.62352244684, + "ttft_p99_ms": 4392.5506810807565, + "tpot_p50_ms": 23.16868536492179, + "tpot_p95_ms": 32.49832417950609, + "tpot_p99_ms": 32.50333228816742, + "e2e_p50_ms": 49589.71686950099, + "e2e_p95_ms": 67762.65414134323, + "e2e_p99_ms": 67792.98369047901, + "itl_p50_ms": 22.87503899424337, + "itl_p95_ms": 24.68757720780559, + "itl_p99_ms": 27.605341025628153, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o4096_a1.json new file mode 100644 index 0000000..4b22b8f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 400, + "failed": 0, + "duration_s": 461.13784259100794, + "request_tps": 0.8674195935699159, + "input_tps": 888.2376638155939, + "output_tps": 3552.9506552623757, + "total_tps": 4441.18831907797, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1903.3365680006682, + "ttft_p95_ms": 3811.413884991634, + "ttft_p99_ms": 4251.1653185045, + "tpot_p50_ms": 21.528175279001097, + "tpot_p95_ms": 23.315381058901853, + "tpot_p99_ms": 23.403718928096588, + "e2e_p50_ms": 90469.03869150265, + "e2e_p95_ms": 96639.56603370316, + "e2e_p99_ms": 96666.38061486068, + "itl_p50_ms": 21.778908499982208, + "itl_p95_ms": 24.338666697440196, + "itl_p99_ms": 26.772771790419934, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o512_a1.json new file mode 100644 index 0000000..d6af508 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c80_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 400, + "failed": 0, + "duration_s": 69.2822109109984, + "request_tps": 5.773487807914353, + "input_tps": 5912.051515304297, + "output_tps": 2956.0257576521485, + "total_tps": 8868.077272956445, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1404.177324497141, + "ttft_p95_ms": 3550.293400152441, + "ttft_p99_ms": 4256.765821308945, + "tpot_p50_ms": 23.07539692855886, + "tpot_p95_ms": 26.923700127783786, + "tpot_p99_ms": 28.125053756603382, + "e2e_p50_ms": 13326.003155001672, + "e2e_p95_ms": 14892.070211340615, + "e2e_p99_ms": 14897.991692002543, + "itl_p50_ms": 22.541024998645298, + "itl_p95_ms": 24.51956669683568, + "itl_p99_ms": 27.620439298043483, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i1024_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i1024_o256_a1.json new file mode 100644 index 0000000..6807d77 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i1024_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 18.495949040006963, + "request_tps": 2.1626357162576255, + "input_tps": 2214.5389734478085, + "output_tps": 553.6347433619521, + "total_tps": 2768.1737168097607, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 440.5348879954545, + "ttft_p95_ms": 3502.821924509044, + "ttft_p99_ms": 3505.5937448414625, + "tpot_p50_ms": 10.14652707840558, + "tpot_p95_ms": 22.57474098336602, + "tpot_p99_ms": 23.09809012723831, + "e2e_p50_ms": 3156.6136774999904, + "e2e_p95_ms": 6075.7029040003545, + "e2e_p99_ms": 6078.332349029661, + "itl_p50_ms": 10.086970491101965, + "itl_p95_ms": 10.408660549001068, + "itl_p99_ms": 11.378483117441656, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i131072_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i131072_o128_a1.json new file mode 100644 index 0000000..9ee9ba7 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i131072_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 380.95681663800497, + "request_tps": 0.1049987774283851, + "input_tps": 13762.399755093293, + "output_tps": 13.439843510833294, + "total_tps": 13775.839598604127, + "mean_input_tokens": 131072.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 40792.213654494844, + "ttft_p95_ms": 75418.06702326574, + "ttft_p99_ms": 76547.42356198141, + "tpot_p50_ms": 274.08619550392257, + "tpot_p95_ms": 467.0889975168508, + "tpot_p99_ms": 530.1239402170978, + "e2e_p50_ms": 77351.3717395108, + "e2e_p95_ms": 77859.17398448946, + "e2e_p99_ms": 82151.91505951836, + "itl_p50_ms": 10.688919486710802, + "itl_p95_ms": 1270.379202303593, + "itl_p99_ms": 1353.2490307680564, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o128_a1.json new file mode 100644 index 0000000..a1c732d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 21.187404705997324, + "request_tps": 1.887914095900456, + "input_tps": 30931.58454723307, + "output_tps": 241.65300427525835, + "total_tps": 31173.237551508326, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 1964.1958890060778, + "ttft_p95_ms": 3285.6321635576023, + "ttft_p99_ms": 3311.663662179344, + "tpot_p50_ms": 17.89670796456699, + "tpot_p95_ms": 32.79861776727342, + "tpot_p99_ms": 35.39528190300389, + "e2e_p50_ms": 4186.461981502362, + "e2e_p95_ms": 4819.31074025124, + "e2e_p99_ms": 4820.3892862127395, + "itl_p50_ms": 12.01575299637625, + "itl_p95_ms": 12.53046725760214, + "itl_p99_ms": 338.20250410339213, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o256_a1.json new file mode 100644 index 0000000..ed3c807 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 25.08967021299759, + "request_tps": 1.5942816171126148, + "input_tps": 26120.71001477308, + "output_tps": 408.1360939808294, + "total_tps": 26528.84610875391, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 1583.4694790100912, + "ttft_p95_ms": 3495.4268617620983, + "ttft_p99_ms": 3574.4406940348563, + "tpot_p50_ms": 12.640562578470801, + "tpot_p95_ms": 19.0468859096262, + "tpot_p99_ms": 19.132231138046265, + "e2e_p50_ms": 4888.153373496607, + "e2e_p95_ms": 6651.8559770600405, + "e2e_p99_ms": 6667.568836012651, + "itl_p50_ms": 11.900605997652747, + "itl_p95_ms": 12.416587599727789, + "itl_p99_ms": 16.99327580136015, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o512_a1.json new file mode 100644 index 0000000..7ea4114 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i16384_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 32.75996512599522, + "request_tps": 1.221002520795108, + "input_tps": 20004.90530070705, + "output_tps": 625.1532906470953, + "total_tps": 20630.058591354147, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 767.6595185039332, + "ttft_p95_ms": 1176.9666902473525, + "ttft_p99_ms": 1830.5755027475243, + "tpot_p50_ms": 11.52257645988111, + "tpot_p95_ms": 13.118042319375098, + "tpot_p99_ms": 13.18724245253204, + "e2e_p50_ms": 6451.215492495976, + "e2e_p95_ms": 7027.763503049937, + "e2e_p99_ms": 7034.832992453594, + "itl_p50_ms": 10.443512990605086, + "itl_p95_ms": 12.125153494707774, + "itl_p99_ms": 12.808226600463968, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i4096_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i4096_o128_a1.json new file mode 100644 index 0000000..9fa9e50 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i4096_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 13.312623847989016, + "request_tps": 3.004666882858133, + "input_tps": 12307.115552186913, + "output_tps": 384.59736100584104, + "total_tps": 12691.712913192754, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 910.605711498647, + "ttft_p95_ms": 1522.6708843554661, + "ttft_p99_ms": 1526.5052215589094, + "tpot_p50_ms": 12.926063618135059, + "tpot_p95_ms": 19.92139204688953, + "tpot_p99_ms": 20.694775890148065, + "e2e_p50_ms": 2659.4583239930216, + "e2e_p95_ms": 3033.2144388041343, + "e2e_p99_ms": 3039.971140041598, + "itl_p50_ms": 10.341610010073055, + "itl_p95_ms": 12.366353905235883, + "itl_p99_ms": 17.325819017132748, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o1024_a1.json new file mode 100644 index 0000000..d9595a1 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 251.80086339799163, + "request_tps": 0.15885569040633815, + "input_tps": 10410.766526469777, + "output_tps": 162.66822697609027, + "total_tps": 10573.434753445867, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 17590.520314995956, + "ttft_p95_ms": 38182.11513094574, + "ttft_p99_ms": 42436.39091802004, + "tpot_p50_ms": 30.034462025418435, + "tpot_p95_ms": 44.99948253655721, + "tpot_p99_ms": 45.840643626068875, + "e2e_p50_ms": 50083.116945505026, + "e2e_p95_ms": 56164.22145070171, + "e2e_p99_ms": 63398.186792482884, + "itl_p50_ms": 12.246457001310773, + "itl_p95_ms": 12.86962560552638, + "itl_p99_ms": 846.2615448146241, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o128_a1.json new file mode 100644 index 0000000..746dc36 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 198.8860901500011, + "request_tps": 0.20112014857264152, + "input_tps": 13180.610056856634, + "output_tps": 25.743379017298114, + "total_tps": 13206.353435873933, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 18747.548280000046, + "ttft_p95_ms": 37954.907318703044, + "ttft_p99_ms": 40759.29950434613, + "tpot_p50_ms": 156.92021938971945, + "tpot_p95_ms": 258.3804646850536, + "tpot_p99_ms": 286.1677625214699, + "e2e_p50_ms": 39171.36516800383, + "e2e_p95_ms": 45190.92419464505, + "e2e_p99_ms": 50389.63774079078, + "itl_p50_ms": 12.328808006714098, + "itl_p95_ms": 904.5139313006074, + "itl_p99_ms": 1091.3648403182742, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o256_a1.json new file mode 100644 index 0000000..f15513a --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 212.1054012759996, + "request_tps": 0.18858548513788426, + "input_tps": 12359.138353996383, + "output_tps": 48.27788419529837, + "total_tps": 12407.416238191681, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 17527.240303992585, + "ttft_p95_ms": 40766.78363170649, + "ttft_p99_ms": 45234.88084075011, + "tpot_p50_ms": 84.98415598624692, + "tpot_p95_ms": 150.79124716706013, + "tpot_p99_ms": 159.6644718746755, + "e2e_p50_ms": 40714.015546502196, + "e2e_p95_ms": 54715.512818601535, + "e2e_p99_ms": 68239.62552806508, + "itl_p50_ms": 12.285736011108384, + "itl_p95_ms": 823.8078383044922, + "itl_p99_ms": 1058.4284620790413, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o512_a1.json new file mode 100644 index 0000000..831c1ae --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c8_i65536_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 221.21875818599074, + "request_tps": 0.1808164928146365, + "input_tps": 11849.989673100017, + "output_tps": 92.57804432109388, + "total_tps": 11942.56771742111, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 19337.339771998813, + "ttft_p95_ms": 38204.29200109792, + "ttft_p99_ms": 42388.65437057044, + "tpot_p50_ms": 48.08991940802787, + "tpot_p95_ms": 75.00450040077983, + "tpot_p99_ms": 78.97179943347363, + "e2e_p50_ms": 44083.91553600086, + "e2e_p95_ms": 48563.43825120639, + "e2e_p99_ms": 51796.121535643615, + "itl_p50_ms": 12.262899006600492, + "itl_p95_ms": 13.326331894495494, + "itl_p99_ms": 956.881430233477, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o1024_a1.json new file mode 100644 index 0000000..a1f60ce --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 480, + "failed": 0, + "duration_s": 131.60086494100688, + "request_tps": 3.647392440886852, + "input_tps": 3734.9298594681363, + "output_tps": 3734.9298594681363, + "total_tps": 7469.859718936273, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1641.8105130069307, + "ttft_p95_ms": 2725.6738467483956, + "ttft_p99_ms": 2848.544350732118, + "tpot_p50_ms": 24.315747637830274, + "tpot_p95_ms": 25.91790127140095, + "tpot_p99_ms": 26.00329389268108, + "e2e_p50_ms": 26516.128896502778, + "e2e_p95_ms": 27687.729706803657, + "e2e_p99_ms": 27709.92693845037, + "itl_p50_ms": 23.818820001906715, + "itl_p95_ms": 26.684747799299657, + "itl_p99_ms": 30.725880123209194, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o4096_a1.json new file mode 100644 index 0000000..245e6c0 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 480, + "failed": 0, + "duration_s": 509.60229949399945, + "request_tps": 0.9419109773967023, + "input_tps": 964.5168408542231, + "output_tps": 3858.0673634168925, + "total_tps": 4822.584204271116, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 2537.5389005057514, + "ttft_p95_ms": 4154.587076739699, + "ttft_p99_ms": 4880.831743057352, + "tpot_p50_ms": 24.489344787912636, + "tpot_p95_ms": 25.11027214580131, + "tpot_p99_ms": 25.309028622957065, + "e2e_p50_ms": 102768.33627750602, + "e2e_p95_ms": 104733.21360939881, + "e2e_p99_ms": 104743.7613113386, + "itl_p50_ms": 24.15896699676523, + "itl_p95_ms": 25.35902190429624, + "itl_p99_ms": 28.550959882268216, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o512_a1.json new file mode 100644 index 0000000..032e58e --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp4_dp2/metrics/vllm_c96_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 480, + "failed": 0, + "duration_s": 75.0480901490082, + "request_tps": 6.3958989368944446, + "input_tps": 6549.400511379911, + "output_tps": 3274.7002556899556, + "total_tps": 9824.100767069867, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1664.6413179987576, + "ttft_p95_ms": 4356.379911445401, + "ttft_p99_ms": 4993.593463639991, + "tpot_p50_ms": 25.16758154697011, + "tpot_p95_ms": 30.006507990595942, + "tpot_p99_ms": 31.581847599314614, + "e2e_p50_ms": 14630.724068490963, + "e2e_p95_ms": 17554.996255355218, + "e2e_p99_ms": 17564.006839842623, + "itl_p50_ms": 23.962745995959267, + "itl_p95_ms": 26.025201004813425, + "itl_p99_ms": 29.676723000011375, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o1024_a1.json new file mode 100644 index 0000000..72249ef --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 61.41648891198565, + "request_tps": 1.3025817889825302, + "input_tps": 1333.843751918111, + "output_tps": 1333.843751918111, + "total_tps": 2667.687503836222, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 362.0326555101201, + "ttft_p95_ms": 606.7866882352973, + "ttft_p99_ms": 607.9597775882576, + "tpot_p50_ms": 11.630160520515538, + "tpot_p95_ms": 11.889058939988779, + "tpot_p99_ms": 11.943324029066188, + "e2e_p50_ms": 12311.110387017834, + "e2e_p95_ms": 12538.315513329871, + "e2e_p99_ms": 12539.58258541039, + "itl_p50_ms": 11.51326100807637, + "itl_p95_ms": 11.765323003055528, + "itl_p99_ms": 11.965371266705917, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o128_a1.json new file mode 100644 index 0000000..e66748f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 52.02482991799479, + "request_tps": 1.5377272761122265, + "input_tps": 1574.63273073892, + "output_tps": 196.829091342365, + "total_tps": 1771.461822081285, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 1236.4054954814492, + "ttft_p95_ms": 35677.24587905686, + "ttft_p99_ms": 35680.42643753841, + "tpot_p50_ms": 15.189849878005925, + "tpot_p95_ms": 63.83978625250509, + "tpot_p99_ms": 164.92346879194884, + "e2e_p50_ms": 2717.587512001046, + "e2e_p95_ms": 37134.60250555654, + "e2e_p99_ms": 37136.07944580115, + "itl_p50_ms": 11.540331004653126, + "itl_p95_ms": 11.923325425595976, + "itl_p99_ms": 14.953987974149642, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o2048_a1.json new file mode 100644 index 0000000..c0fe994 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 123.58838067599572, + "request_tps": 0.6473100429216823, + "input_tps": 662.8454839518026, + "output_tps": 1325.6909679036053, + "total_tps": 1988.536451855408, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 503.70064900198486, + "ttft_p95_ms": 3332.6880413966137, + "ttft_p99_ms": 3334.5142474136082, + "tpot_p50_ms": 11.566837329992605, + "tpot_p95_ms": 11.80489444616316, + "tpot_p99_ms": 12.187286488214971, + "e2e_p50_ms": 24182.123156497255, + "e2e_p95_ms": 27096.394883874746, + "e2e_p99_ms": 27099.026713256026, + "itl_p50_ms": 11.55229500727728, + "itl_p95_ms": 11.839765001786873, + "itl_p99_ms": 12.08073129178956, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o256_a1.json new file mode 100644 index 0000000..af35ed2 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 61.86757407398545, + "request_tps": 1.2930844824193455, + "input_tps": 1324.1185099974098, + "output_tps": 331.02962749935244, + "total_tps": 1655.1481374967623, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 392.70260799094103, + "ttft_p95_ms": 27561.146233863838, + "ttft_p99_ms": 27563.441789576318, + "tpot_p50_ms": 12.039197992143112, + "tpot_p95_ms": 70.91977109907734, + "tpot_p99_ms": 90.06260683440567, + "e2e_p50_ms": 3515.55855650804, + "e2e_p95_ms": 30491.65073121112, + "e2e_p99_ms": 30493.552179136605, + "itl_p50_ms": 11.53830900148023, + "itl_p95_ms": 11.843036256323103, + "itl_p99_ms": 12.743973096075921, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o4096_a1.json new file mode 100644 index 0000000..dff7d12 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 238.9891159539984, + "request_tps": 0.3347432776620619, + "input_tps": 342.7771163259514, + "output_tps": 1371.1084653038056, + "total_tps": 1713.885581629757, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 407.33929100679234, + "ttft_p95_ms": 588.6433666979428, + "ttft_p99_ms": 591.8217892898247, + "tpot_p50_ms": 11.581026416729486, + "tpot_p95_ms": 11.690784101489584, + "tpot_p99_ms": 11.69744405043216, + "e2e_p50_ms": 47807.511286999215, + "e2e_p95_ms": 48180.03251795162, + "e2e_p99_ms": 48181.1768043015, + "itl_p50_ms": 11.548227979801595, + "itl_p95_ms": 11.81682851165533, + "itl_p99_ms": 11.96142429544125, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o512_a1.json new file mode 100644 index 0000000..a9a2c0d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 32.14737898000749, + "request_tps": 2.488538802797956, + "input_tps": 2548.2637340651067, + "output_tps": 1274.1318670325534, + "total_tps": 3822.39560109766, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 413.3373150107218, + "ttft_p95_ms": 656.7111001349986, + "ttft_p99_ms": 658.238969738304, + "tpot_p50_ms": 11.758502294494425, + "tpot_p95_ms": 12.275672713020104, + "tpot_p99_ms": 12.393381951049646, + "e2e_p50_ms": 6405.572134506656, + "e2e_p95_ms": 6744.097330357181, + "e2e_p99_ms": 6745.546151806775, + "itl_p50_ms": 11.505689995829016, + "itl_p95_ms": 11.79091099766083, + "itl_p99_ms": 12.107418500818312, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i131072_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i131072_o128_a1.json new file mode 100644 index 0000000..7b490b7 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i131072_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 1061.9543217770115, + "request_tps": 0.07533280703272881, + "input_tps": 9874.02168339383, + "output_tps": 9.642599300189287, + "total_tps": 9883.664282694019, + "mean_input_tokens": 131072.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 109194.33135900181, + "ttft_p95_ms": 129901.68803883859, + "ttft_p99_ms": 173369.75827682696, + "tpot_p50_ms": 860.7348704526814, + "tpot_p95_ms": 863.007278436844, + "tpot_p99_ms": 877.5824045137604, + "e2e_p50_ms": 218323.68925350602, + "e2e_p95_ms": 239327.92834405843, + "e2e_p99_ms": 282901.98067860294, + "itl_p50_ms": 857.2086700005457, + "itl_p95_ms": 1084.6890412460198, + "itl_p99_ms": 1101.4063146195258, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o1024_a1.json new file mode 100644 index 0000000..11cc5a3 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 113.21908437900129, + "request_tps": 0.7065946561818122, + "input_tps": 11576.84684688281, + "output_tps": 723.5529279301757, + "total_tps": 12300.399774812988, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1313.6420960072428, + "ttft_p95_ms": 15357.72944069031, + "ttft_p99_ms": 15506.10770056053, + "tpot_p50_ms": 15.984088203331268, + "tpot_p95_ms": 27.630180807318535, + "tpot_p99_ms": 30.014112934037232, + "e2e_p50_ms": 22529.161178987124, + "e2e_p95_ms": 37872.18460994045, + "e2e_p99_ms": 38033.13407269189, + "itl_p50_ms": 11.689694976666942, + "itl_p95_ms": 12.091477410285734, + "itl_p99_ms": 100.4964998154901, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o128_a1.json new file mode 100644 index 0000000..9b82e9d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 86.4904059410037, + "request_tps": 0.924958082108773, + "input_tps": 15154.513217270136, + "output_tps": 118.39463450992294, + "total_tps": 15272.907851780057, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 7593.278043495957, + "ttft_p95_ms": 12824.893658250221, + "ttft_p99_ms": 14977.241600849084, + "tpot_p50_ms": 78.05374694479846, + "tpot_p95_ms": 140.04996209068491, + "tpot_p99_ms": 159.51224197268965, + "e2e_p50_ms": 17123.760913003935, + "e2e_p95_ms": 26093.928836856503, + "e2e_p99_ms": 26102.272943213466, + "itl_p50_ms": 11.707582991220988, + "itl_p95_ms": 648.0138543454814, + "itl_p99_ms": 666.6313856895431, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o2048_a1.json new file mode 100644 index 0000000..575eb41 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 125.30134890798945, + "request_tps": 0.6384608042707116, + "input_tps": 10460.541817171339, + "output_tps": 1307.5677271464174, + "total_tps": 11768.109544317756, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 779.641143992194, + "ttft_p95_ms": 1110.0852024552294, + "ttft_p99_ms": 1200.1118386295275, + "tpot_p50_ms": 11.868535543230175, + "tpot_p95_ms": 12.105919975430762, + "tpot_p99_ms": 12.135692873646816, + "e2e_p50_ms": 25057.22738750046, + "e2e_p95_ms": 25164.101334703446, + "e2e_p99_ms": 25175.635717739642, + "itl_p50_ms": 11.723723000613973, + "itl_p95_ms": 12.08025178639218, + "itl_p99_ms": 12.76948868646286, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o256_a1.json new file mode 100644 index 0000000..d548f55 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 20.272567973006517, + "request_tps": 3.946219349542801, + "input_tps": 64654.85782290925, + "output_tps": 1010.2321534829571, + "total_tps": 65665.0899763922, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 732.9621660028351, + "ttft_p95_ms": 1059.6370970582939, + "ttft_p99_ms": 1330.065922466165, + "tpot_p50_ms": 12.931638015616757, + "tpot_p95_ms": 14.641560330985836, + "tpot_p99_ms": 15.307604808072721, + "e2e_p50_ms": 3975.734623003518, + "e2e_p95_ms": 4291.209159711434, + "e2e_p99_ms": 4300.588087121723, + "itl_p50_ms": 11.725543983629905, + "itl_p95_ms": 12.27384575031465, + "itl_p99_ms": 21.68184940965146, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o512_a1.json new file mode 100644 index 0000000..ff89253 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i16384_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 130.2362937349826, + "request_tps": 0.6142680945972843, + "input_tps": 10064.168461881905, + "output_tps": 314.50526443380954, + "total_tps": 10378.673726315716, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 8992.384835495614, + "ttft_p95_ms": 14412.819043596388, + "ttft_p99_ms": 18462.384867458015, + "tpot_p50_ms": 34.2774762993976, + "tpot_p95_ms": 47.27433437131593, + "tpot_p99_ms": 48.483385741755086, + "e2e_p50_ms": 26175.92193999735, + "e2e_p95_ms": 32332.358848227883, + "e2e_p99_ms": 36386.61295437544, + "itl_p50_ms": 11.672908993205056, + "itl_p95_ms": 13.615285683772521, + "itl_p99_ms": 648.6071743268986, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o1024_a1.json new file mode 100644 index 0000000..f46b77b --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 62.43282029798138, + "request_tps": 1.2813773207452974, + "input_tps": 5248.521505772738, + "output_tps": 1312.1303764431846, + "total_tps": 6560.651882215922, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 563.8070285058348, + "ttft_p95_ms": 777.6129835590837, + "ttft_p99_ms": 779.2977576740668, + "tpot_p50_ms": 11.56457677565038, + "tpot_p95_ms": 12.043608254741104, + "tpot_p99_ms": 12.209488064032353, + "e2e_p50_ms": 12494.566029999987, + "e2e_p95_ms": 12608.202403750329, + "e2e_p99_ms": 12608.88043828163, + "itl_p50_ms": 11.578381003346294, + "itl_p95_ms": 11.835356391384266, + "itl_p99_ms": 12.015851708129048, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o128_a1.json new file mode 100644 index 0000000..786ad96 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 31.652223498997046, + "request_tps": 2.527468567967585, + "input_tps": 10352.511254395227, + "output_tps": 323.51597669985085, + "total_tps": 10676.027231095079, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 2794.544570497237, + "ttft_p95_ms": 4019.1005029933863, + "ttft_p99_ms": 4515.16735956771, + "tpot_p50_ms": 28.070199511753664, + "tpot_p95_ms": 46.8207419802839, + "tpot_p99_ms": 47.12049638943226, + "e2e_p50_ms": 6348.330602006172, + "e2e_p95_ms": 6970.749582740245, + "e2e_p99_ms": 7275.766055090349, + "itl_p50_ms": 11.53999000962358, + "itl_p95_ms": 12.58592710510129, + "itl_p99_ms": 611.4497984829359, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o2048_a1.json new file mode 100644 index 0000000..664c4e8 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 121.8976368480071, + "request_tps": 0.6562883585655658, + "input_tps": 2688.1571166845574, + "output_tps": 1344.0785583422787, + "total_tps": 4032.235675026836, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 452.4103995063342, + "ttft_p95_ms": 869.4939421853633, + "ttft_p99_ms": 873.1171497041942, + "tpot_p50_ms": 11.663301339766598, + "tpot_p95_ms": 11.80714733549408, + "tpot_p99_ms": 11.878064581100528, + "e2e_p50_ms": 24396.55241399305, + "e2e_p95_ms": 24617.466119487653, + "e2e_p99_ms": 24621.38425450714, + "itl_p50_ms": 11.577293000300415, + "itl_p95_ms": 11.859655751322862, + "itl_p99_ms": 12.521442596334962, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o256_a1.json new file mode 100644 index 0000000..4ceb101 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 17.724754639988532, + "request_tps": 4.513461631762919, + "input_tps": 18487.138843700915, + "output_tps": 1155.4461777313072, + "total_tps": 19642.58502143222, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 584.0600009978516, + "ttft_p95_ms": 737.9011562516098, + "ttft_p99_ms": 741.6909766758907, + "tpot_p50_ms": 11.493453029349135, + "tpot_p95_ms": 13.185771285709194, + "tpot_p99_ms": 13.456647717705515, + "e2e_p50_ms": 3522.1422600006917, + "e2e_p95_ms": 3669.390212147846, + "e2e_p99_ms": 3672.551417663053, + "itl_p50_ms": 11.582269973587245, + "itl_p95_ms": 11.904317999142222, + "itl_p99_ms": 12.9585687990766, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o4096_a1.json new file mode 100644 index 0000000..011aa2e --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 239.23693925398402, + "request_tps": 0.3343965202425058, + "input_tps": 1369.6881469133039, + "output_tps": 1369.6881469133039, + "total_tps": 2739.3762938266077, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 535.376151005039, + "ttft_p95_ms": 764.6692137466744, + "ttft_p99_ms": 770.1709890385973, + "tpot_p50_ms": 11.550237434308961, + "tpot_p95_ms": 11.648775638909902, + "tpot_p99_ms": 11.68880080481187, + "e2e_p50_ms": 47866.3036479993, + "e2e_p95_ms": 48047.369737493864, + "e2e_p99_ms": 48052.40901362762, + "itl_p50_ms": 11.535027006175369, + "itl_p95_ms": 11.756750987842679, + "itl_p99_ms": 11.894877816084774, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o512_a1.json new file mode 100644 index 0000000..4eebaf9 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i4096_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 32.73595824901713, + "request_tps": 2.4437958831524944, + "input_tps": 10009.787937392617, + "output_tps": 1251.2234921740771, + "total_tps": 11261.011429566694, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 580.2664404909592, + "ttft_p95_ms": 736.464851902565, + "ttft_p99_ms": 739.5049455872504, + "tpot_p50_ms": 11.585709298447135, + "tpot_p95_ms": 12.479319499847486, + "tpot_p99_ms": 12.621792366471547, + "e2e_p50_ms": 6533.585163517273, + "e2e_p95_ms": 6650.0653961644275, + "e2e_p99_ms": 6653.992265443085, + "itl_p50_ms": 11.581516009755433, + "itl_p95_ms": 11.963144004403148, + "itl_p99_ms": 12.631978923745926, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o1024_a1.json new file mode 100644 index 0000000..d8852ab --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 500.2035044890072, + "request_tps": 0.15993490505774363, + "input_tps": 10481.493937864287, + "output_tps": 163.77334277912948, + "total_tps": 10645.267280643417, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 30243.93685850373, + "ttft_p95_ms": 42969.91652948925, + "ttft_p99_ms": 61614.389155033525, + "tpot_p50_ms": 73.36121226490721, + "tpot_p95_ms": 91.33674583832028, + "tpot_p99_ms": 96.64777304192485, + "e2e_p50_ms": 105178.19468899688, + "e2e_p95_ms": 113684.78196981303, + "e2e_p99_ms": 132258.0071013729, + "itl_p50_ms": 12.48324201151263, + "itl_p95_ms": 713.5221524600638, + "itl_p99_ms": 825.3858603240224, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o128_a1.json new file mode 100644 index 0000000..c9d2c5d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 467.49062519401195, + "request_tps": 0.17112642626106017, + "input_tps": 11214.94147144484, + "output_tps": 21.904182561415702, + "total_tps": 11236.845654006254, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 28092.216243006987, + "ttft_p95_ms": 65609.9809892679, + "ttft_p99_ms": 84234.906140567, + "tpot_p50_ms": 529.0584835078563, + "tpot_p95_ms": 689.1646421382568, + "tpot_p99_ms": 694.2038618913791, + "e2e_p50_ms": 94517.5517155003, + "e2e_p95_ms": 132926.59808922032, + "e2e_p99_ms": 151589.5588841254, + "itl_p50_ms": 700.4347535112174, + "itl_p95_ms": 837.8521792372339, + "itl_p99_ms": 848.7232003509416, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o256_a1.json new file mode 100644 index 0000000..a92bf47 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 452.2059691059985, + "request_tps": 0.17691053516643815, + "input_tps": 11594.00883266769, + "output_tps": 45.289097002608166, + "total_tps": 11639.2979296703, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 27964.375940995524, + "ttft_p95_ms": 42944.88978464359, + "ttft_p99_ms": 61584.43124415524, + "tpot_p50_ms": 264.55735394904644, + "tpot_p95_ms": 328.878599924461, + "tpot_p99_ms": 350.21582279077904, + "e2e_p50_ms": 95521.30952950392, + "e2e_p95_ms": 106008.97152810794, + "e2e_p99_ms": 121054.057937187, + "itl_p50_ms": 12.545981488074176, + "itl_p95_ms": 818.9031008194432, + "itl_p99_ms": 841.0442970949224, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o512_a1.json new file mode 100644 index 0000000..784ff38 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c16_i65536_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 80, + "failed": 0, + "duration_s": 468.2541085879784, + "request_tps": 0.1708474064247727, + "input_tps": 11196.655627453903, + "output_tps": 87.47387208948362, + "total_tps": 11284.129499543387, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 28707.305190488114, + "ttft_p95_ms": 42941.14238748442, + "ttft_p99_ms": 61583.86625094918, + "tpot_p50_ms": 138.2095831995791, + "tpot_p95_ms": 170.5406067458202, + "tpot_p99_ms": 181.07553997671428, + "e2e_p50_ms": 98761.97971450165, + "e2e_p95_ms": 112307.20164615194, + "e2e_p99_ms": 130892.13807505231, + "itl_p50_ms": 12.49102050496731, + "itl_p95_ms": 781.4320894540288, + "itl_p99_ms": 836.5383420526632, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o128_a1.json new file mode 100644 index 0000000..9837e60 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 4.607907787984004, + "request_tps": 1.0850911585163339, + "input_tps": 1111.1333463207259, + "output_tps": 138.89166829009073, + "total_tps": 1250.0250146108165, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 68.60028099617921, + "ttft_p95_ms": 93.32109419628976, + "ttft_p99_ms": 98.21370284073053, + "tpot_p50_ms": 6.665402291183118, + "tpot_p95_ms": 6.668919995271301, + "tpot_p99_ms": 6.669149680904619, + "e2e_p50_ms": 914.933813008247, + "e2e_p95_ms": 940.2739335957449, + "e2e_p99_ms": 945.1957123156171, + "itl_p50_ms": 6.641707994276658, + "itl_p95_ms": 6.847297985223121, + "itl_p99_ms": 7.190719614736732, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o256_a1.json new file mode 100644 index 0000000..3b0090f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i1024_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 8.878280922974227, + "request_tps": 0.5631720874095746, + "input_tps": 576.6882175074044, + "output_tps": 144.1720543768511, + "total_tps": 720.8602718842554, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 71.51998899644241, + "ttft_p95_ms": 91.57086079358123, + "ttft_p99_ms": 95.35931215505116, + "tpot_p50_ms": 6.661176898028227, + "tpot_p95_ms": 6.664835591321154, + "tpot_p99_ms": 6.665386227228879, + "e2e_p50_ms": 1771.2285799789242, + "e2e_p95_ms": 1790.1660829898901, + "e2e_p99_ms": 1793.9073757897131, + "itl_p50_ms": 6.639207014814019, + "itl_p95_ms": 6.857857707655057, + "itl_p99_ms": 7.324532314087264, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i131072_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i131072_o128_a1.json new file mode 100644 index 0000000..6fbbc43 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i131072_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 62.04160683800001, + "request_tps": 0.08059107838802039, + "input_tps": 10563.233826474609, + "output_tps": 10.31565803366661, + "total_tps": 10573.549484508274, + "mean_input_tokens": 131072.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 14315.946837014053, + "ttft_p95_ms": 14424.979343393352, + "ttft_p99_ms": 14444.436821474228, + "tpot_p50_ms": 6.151024771691117, + "tpot_p95_ms": 6.183124977891795, + "tpot_p99_ms": 6.187243395545484, + "e2e_p50_ms": 15091.257207997842, + "e2e_p95_ms": 15205.586195399519, + "e2e_p99_ms": 15225.502308679279, + "itl_p50_ms": 6.912430500960909, + "itl_p95_ms": 7.0415690206573345, + "itl_p99_ms": 7.134565238084179, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i16384_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i16384_o512_a1.json new file mode 100644 index 0000000..d51cf93 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i16384_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 18.15804166198359, + "request_tps": 0.2753600907562737, + "input_tps": 4511.4997269507885, + "output_tps": 140.98436646721214, + "total_tps": 4652.484093418, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 186.52946400106885, + "ttft_p95_ms": 232.34625218901783, + "ttft_p99_ms": 237.30703042820096, + "tpot_p50_ms": 6.7337785733917475, + "tpot_p95_ms": 6.740160819146246, + "tpot_p99_ms": 6.740686006462676, + "e2e_p50_ms": 3627.490315004252, + "e2e_p95_ms": 3669.8924877971876, + "e2e_p99_ms": 3676.717489558505, + "itl_p50_ms": 6.754819012712687, + "itl_p95_ms": 6.876269215717912, + "itl_p99_ms": 7.019366837339476, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o1024_a1.json new file mode 100644 index 0000000..bf4b5c5 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 60.06458672598819, + "request_tps": 0.08324372600463838, + "input_tps": 5455.460827439981, + "output_tps": 85.2415754287497, + "total_tps": 5540.70240286873, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 6200.891699991189, + "ttft_p95_ms": 6272.143243986648, + "ttft_p99_ms": 6285.5985119822435, + "tpot_p50_ms": 6.7929081173096115, + "tpot_p95_ms": 6.796755083673428, + "tpot_p99_ms": 6.797365722498417, + "e2e_p50_ms": 13150.848732999293, + "e2e_p95_ms": 13210.16350100399, + "e2e_p99_ms": 13220.45055940398, + "itl_p50_ms": 6.836396001745015, + "itl_p95_ms": 6.958074710564688, + "itl_p99_ms": 7.039473712211475, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o128_a1.json new file mode 100644 index 0000000..687a49e --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 29.374553603003733, + "request_tps": 0.170215352633945, + "input_tps": 11155.23335021822, + "output_tps": 21.78756513714496, + "total_tps": 11177.020915355364, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 6167.725449980935, + "ttft_p95_ms": 6247.644395998213, + "ttft_p99_ms": 6256.791441601235, + "tpot_p50_ms": 6.4021568109673685, + "tpot_p95_ms": 6.590872122924481, + "tpot_p99_ms": 6.591545397903419, + "e2e_p50_ms": 7004.3386799807195, + "e2e_p95_ms": 7057.308506994741, + "e2e_p99_ms": 7065.896558194654, + "itl_p50_ms": 6.835169973783195, + "itl_p95_ms": 6.938054598867893, + "itl_p99_ms": 7.164961716625845, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o256_a1.json new file mode 100644 index 0000000..b35a085 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 33.741494465008145, + "request_tps": 0.14818549324142372, + "input_tps": 9711.484485069945, + "output_tps": 37.93548626980447, + "total_tps": 9749.419971339748, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 6187.758629996097, + "ttft_p95_ms": 6245.184306602459, + "ttft_p99_ms": 6254.763505321462, + "tpot_p50_ms": 6.614399407921797, + "tpot_p95_ms": 6.6246079560597515, + "tpot_p99_ms": 6.626020995101105, + "e2e_p50_ms": 7877.484064985765, + "e2e_p95_ms": 7930.040321213892, + "e2e_p99_ms": 7939.256353051169, + "itl_p50_ms": 6.8363494938239455, + "itl_p95_ms": 6.942608255485538, + "itl_p99_ms": 7.0092535417643385, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o512_a1.json new file mode 100644 index 0000000..e36eeca --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c1_i65536_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 5, + "failed": 0, + "duration_s": 42.45334394200472, + "request_tps": 0.11777635247839305, + "input_tps": 7718.591036023967, + "output_tps": 60.30149246893724, + "total_tps": 7778.892528492904, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 6204.497470986098, + "ttft_p95_ms": 6222.823049785802, + "ttft_p99_ms": 6224.350896343822, + "tpot_p50_ms": 6.734576565566621, + "tpot_p95_ms": 6.741488577286328, + "tpot_p99_ms": 6.742628805848284, + "e2e_p50_ms": 9645.783593994565, + "e2e_p95_ms": 9660.80251399544, + "e2e_p99_ms": 9663.666046793805, + "itl_p50_ms": 6.839831999968737, + "itl_p95_ms": 6.9495914547587745, + "itl_p99_ms": 7.086883967567698, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o1024_a1.json new file mode 100644 index 0000000..04c1d1c --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 80.85768987398478, + "request_tps": 1.9787851996434358, + "input_tps": 2026.2760444348783, + "output_tps": 2026.2760444348783, + "total_tps": 4052.5520888697565, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 601.8352310056798, + "ttft_p95_ms": 875.1060022099409, + "ttft_p99_ms": 879.6998479493777, + "tpot_p50_ms": 15.16937012902533, + "tpot_p95_ms": 15.621438565148944, + "tpot_p99_ms": 15.709627801430935, + "e2e_p50_ms": 16193.191637503332, + "e2e_p95_ms": 16394.8902862423, + "e2e_p99_ms": 16400.64757927088, + "itl_p50_ms": 15.000293991761282, + "itl_p95_ms": 15.395940997404976, + "itl_p99_ms": 15.808226090157401, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o2048_a1.json new file mode 100644 index 0000000..121ff79 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 158.5647476890008, + "request_tps": 1.0090515220559253, + "input_tps": 1033.2687585852675, + "output_tps": 2066.537517170535, + "total_tps": 3099.8062757558023, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 669.473130008555, + "ttft_p95_ms": 865.0462039600825, + "ttft_p99_ms": 869.1866847060737, + "tpot_p50_ms": 15.184576202977428, + "tpot_p95_ms": 15.339674701513589, + "tpot_p99_ms": 15.58220753159747, + "e2e_p50_ms": 31597.606307492242, + "e2e_p95_ms": 32205.554889343333, + "e2e_p99_ms": 32214.89342865796, + "itl_p50_ms": 15.059143508551642, + "itl_p95_ms": 15.587451988540124, + "itl_p99_ms": 16.146498053567488, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o4096_a1.json new file mode 100644 index 0000000..c3b2076 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 315.3864424380008, + "request_tps": 0.507314134251199, + "input_tps": 519.4896734732278, + "output_tps": 2077.958693892911, + "total_tps": 2597.448367366139, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 827.0409834949533, + "ttft_p95_ms": 2127.0444475507247, + "ttft_p99_ms": 2149.356553926773, + "tpot_p50_ms": 15.163013353477309, + "tpot_p95_ms": 15.352130105484576, + "tpot_p99_ms": 15.488373952998492, + "e2e_p50_ms": 63035.64160001406, + "e2e_p95_ms": 63718.42900119955, + "e2e_p99_ms": 63728.90193682426, + "itl_p50_ms": 15.009783499408513, + "itl_p95_ms": 15.42943195818225, + "itl_p99_ms": 15.688629885262342, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o512_a1.json new file mode 100644 index 0000000..5ac794c --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 51.012810627988074, + "request_tps": 3.1364670566145265, + "input_tps": 3211.742265973275, + "output_tps": 1605.8711329866376, + "total_tps": 4817.613398959913, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1028.030455505359, + "ttft_p95_ms": 1825.9766799485078, + "ttft_p99_ms": 2148.851445445325, + "tpot_p50_ms": 17.623297716239186, + "tpot_p95_ms": 19.62757935753655, + "tpot_p99_ms": 20.25833365699475, + "e2e_p50_ms": 10080.087392503629, + "e2e_p95_ms": 11066.638421798416, + "e2e_p99_ms": 11499.76811571978, + "itl_p50_ms": 16.898524991120212, + "itl_p95_ms": 17.623670147440862, + "itl_p99_ms": 18.77276543091283, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o2048_a1.json new file mode 100644 index 0000000..e76d854 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 262.4147839480138, + "request_tps": 0.6097217450663798, + "input_tps": 9989.681071167566, + "output_tps": 1248.7101338959458, + "total_tps": 11238.391205063512, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 2030.8082264964469, + "ttft_p95_ms": 23697.570598886523, + "ttft_p99_ms": 24982.915203587618, + "tpot_p50_ms": 23.179559257443874, + "tpot_p95_ms": 31.551512422186658, + "tpot_p99_ms": 33.554478466176256, + "e2e_p50_ms": 52189.317357493564, + "e2e_p95_ms": 86702.39336090744, + "e2e_p99_ms": 87993.08046271733, + "itl_p50_ms": 15.3286209970247, + "itl_p95_ms": 15.942659782012925, + "itl_p99_ms": 610.8087138849077, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o256_a1.json new file mode 100644 index 0000000..bcca135 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i16384_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 125.27427309198538, + "request_tps": 1.2771975925377472, + "input_tps": 20925.60535613845, + "output_tps": 326.9625836896633, + "total_tps": 21252.567939828114, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 1955.9702225087676, + "ttft_p95_ms": 25687.426767007855, + "ttft_p99_ms": 26337.214467667218, + "tpot_p50_ms": 71.67799913920189, + "tpot_p95_ms": 145.99830021673134, + "tpot_p99_ms": 162.3011558566975, + "e2e_p50_ms": 24834.69372948457, + "e2e_p95_ms": 61469.757113161904, + "e2e_p99_ms": 62129.46268044296, + "itl_p50_ms": 15.310183982364833, + "itl_p95_ms": 637.7874485973734, + "itl_p99_ms": 661.2853236787487, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o1024_a1.json new file mode 100644 index 0000000..062b6ad --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 103.49840845502331, + "request_tps": 1.5459174917605643, + "input_tps": 6332.078046251271, + "output_tps": 1583.0195115628178, + "total_tps": 7915.097557814089, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 912.6817460055463, + "ttft_p95_ms": 22953.754147146537, + "ttft_p99_ms": 22989.648280964466, + "tpot_p50_ms": 15.027860665688419, + "tpot_p95_ms": 15.874237878655665, + "tpot_p99_ms": 34.52500240562136, + "e2e_p50_ms": 16423.773564994917, + "e2e_p95_ms": 38356.92362101836, + "e2e_p99_ms": 38391.64368835016, + "itl_p50_ms": 15.018527017673478, + "itl_p95_ms": 15.399335196707398, + "itl_p99_ms": 15.800710814073682, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o2048_a1.json new file mode 100644 index 0000000..be495c4 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 158.23376187801477, + "request_tps": 1.0111622077426614, + "input_tps": 4141.720402913941, + "output_tps": 2070.8602014569706, + "total_tps": 6212.580604370912, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 756.7443619918777, + "ttft_p95_ms": 1068.711685812741, + "ttft_p99_ms": 1076.5162225518725, + "tpot_p50_ms": 15.088843866877845, + "tpot_p95_ms": 15.323755214356261, + "tpot_p99_ms": 15.423469172625605, + "e2e_p50_ms": 31670.834326505428, + "e2e_p95_ms": 31954.730625603406, + "e2e_p99_ms": 31962.703760440054, + "itl_p50_ms": 14.991424977779388, + "itl_p95_ms": 15.389906987547874, + "itl_p99_ms": 15.759109035134308, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o256_a1.json new file mode 100644 index 0000000..9835d00 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 46.13867022198974, + "request_tps": 3.467806922700252, + "input_tps": 14204.137155380233, + "output_tps": 887.7585722112645, + "total_tps": 15091.895727591496, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 1084.7839874913916, + "ttft_p95_ms": 6128.039478484425, + "ttft_p99_ms": 6135.151792074903, + "tpot_p50_ms": 28.80620697841925, + "tpot_p95_ms": 44.439678273264136, + "tpot_p99_ms": 49.92359239300004, + "e2e_p50_ms": 9273.995624491363, + "e2e_p95_ms": 15928.954499101383, + "e2e_p99_ms": 15931.015178921225, + "itl_p50_ms": 14.997396006947383, + "itl_p95_ms": 15.599585502059199, + "itl_p99_ms": 613.4823884931393, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o4096_a1.json new file mode 100644 index 0000000..157fe16 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 333.05284896001103, + "request_tps": 0.4804042376446114, + "input_tps": 1967.7357573923282, + "output_tps": 1967.7357573923282, + "total_tps": 3935.4715147846564, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1095.2641390031204, + "ttft_p95_ms": 6705.06853362167, + "ttft_p99_ms": 7314.000860454107, + "tpot_p50_ms": 15.522370325397588, + "tpot_p95_ms": 16.71266772969576, + "tpot_p99_ms": 17.02914903613981, + "e2e_p50_ms": 66306.44073350413, + "e2e_p95_ms": 73349.6555153004, + "e2e_p99_ms": 73959.53097622347, + "itl_p50_ms": 14.9437389918603, + "itl_p95_ms": 15.368272416526452, + "itl_p99_ms": 15.751762358704575, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o512_a1.json new file mode 100644 index 0000000..d8ccb0a --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c32_i4096_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 160, + "failed": 0, + "duration_s": 43.90981093500159, + "request_tps": 3.6438325875928577, + "input_tps": 14925.138278780345, + "output_tps": 1865.6422848475431, + "total_tps": 16790.78056362789, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 799.3582674971549, + "ttft_p95_ms": 1175.5077986585093, + "ttft_p99_ms": 1183.6164445849136, + "tpot_p50_ms": 15.426591234855762, + "tpot_p95_ms": 17.008792841357067, + "tpot_p99_ms": 17.19561321969741, + "e2e_p50_ms": 8660.748706999584, + "e2e_p95_ms": 9361.31468063686, + "e2e_p99_ms": 9369.353665226081, + "itl_p50_ms": 15.0350559852086, + "itl_p95_ms": 17.14442181983031, + "itl_p99_ms": 17.799102468416095, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o1024_a1.json new file mode 100644 index 0000000..ec8fd07 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 94.47741836198838, + "request_tps": 2.5402895650730493, + "input_tps": 2601.2565146348024, + "output_tps": 2601.2565146348024, + "total_tps": 5202.513029269605, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 872.004470002139, + "ttft_p95_ms": 1068.056017033814, + "ttft_p99_ms": 1074.3570806449861, + "tpot_p50_ms": 17.694852847022954, + "tpot_p95_ms": 18.194023427918026, + "tpot_p99_ms": 18.255092446847165, + "e2e_p50_ms": 18907.909675486735, + "e2e_p95_ms": 19168.525980616687, + "e2e_p99_ms": 19178.07574301085, + "itl_p50_ms": 17.358476005028933, + "itl_p95_ms": 17.918874911265448, + "itl_p99_ms": 18.629112805938348, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o2048_a1.json new file mode 100644 index 0000000..ff2248f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 187.52933345900965, + "request_tps": 1.279799781576355, + "input_tps": 1310.5149763341876, + "output_tps": 2621.0299526683752, + "total_tps": 3931.5449290025626, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 996.749361496768, + "ttft_p95_ms": 2177.432288348791, + "ttft_p99_ms": 2548.245648367156, + "tpot_p50_ms": 17.70298705446327, + "tpot_p95_ms": 18.496204302150304, + "tpot_p99_ms": 18.711075803456517, + "e2e_p50_ms": 37275.27416749217, + "e2e_p95_ms": 39046.80715519644, + "e2e_p99_ms": 39530.66943249112, + "itl_p50_ms": 17.395492002833635, + "itl_p95_ms": 17.924149986356497, + "itl_p99_ms": 18.54355659452267, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o4096_a1.json new file mode 100644 index 0000000..437cad0 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 384.594265947002, + "request_tps": 0.6240342648090146, + "input_tps": 639.0110871644309, + "output_tps": 2556.0443486577237, + "total_tps": 3195.0554358221543, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 988.36895050772, + "ttft_p95_ms": 2441.5721719269623, + "ttft_p99_ms": 5020.39980761183, + "tpot_p50_ms": 17.599965752136836, + "tpot_p95_ms": 22.463060006479317, + "tpot_p99_ms": 22.464100362125627, + "e2e_p50_ms": 73127.70336250833, + "e2e_p95_ms": 92961.33970778319, + "e2e_p99_ms": 93251.61461467971, + "itl_p50_ms": 17.415053996955976, + "itl_p95_ms": 17.90137597708963, + "itl_p99_ms": 18.575296591734517, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o512_a1.json new file mode 100644 index 0000000..ef08044 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i1024_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 85.11057490797248, + "request_tps": 2.819861107265517, + "input_tps": 2887.5377738398893, + "output_tps": 1443.7688869199446, + "total_tps": 4331.306660759834, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1176.0342860070523, + "ttft_p95_ms": 32154.562560931667, + "ttft_p99_ms": 32162.17692289414, + "tpot_p50_ms": 18.486646824851704, + "tpot_p95_ms": 74.38846726118646, + "tpot_p99_ms": 74.39255435188375, + "e2e_p50_ms": 11688.053546997253, + "e2e_p95_ms": 41145.71591668209, + "e2e_p99_ms": 41149.52881608595, + "itl_p50_ms": 17.28113798890263, + "itl_p95_ms": 17.941987194353715, + "itl_p99_ms": 19.419972189934924, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o2048_a1.json new file mode 100644 index 0000000..1ff2246 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 206.79596914799185, + "request_tps": 1.1605642072657902, + "input_tps": 4753.670992960677, + "output_tps": 2376.8354964803384, + "total_tps": 7130.506489441016, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1145.1450809981907, + "ttft_p95_ms": 6598.897911023221, + "ttft_p99_ms": 9627.660050078412, + "tpot_p50_ms": 17.8303520544688, + "tpot_p95_ms": 22.353743953683303, + "tpot_p99_ms": 23.511489080978755, + "e2e_p50_ms": 37381.53092398716, + "e2e_p95_ms": 50990.53589172253, + "e2e_p99_ms": 54006.87992766645, + "itl_p50_ms": 17.39579500281252, + "itl_p95_ms": 17.878196993842717, + "itl_p99_ms": 18.62087064306253, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o512_a1.json new file mode 100644 index 0000000..2de9da6 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c48_i4096_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 240, + "failed": 0, + "duration_s": 76.34642335801618, + "request_tps": 3.1435657290001995, + "input_tps": 12876.045225984817, + "output_tps": 1609.5056532481021, + "total_tps": 14485.55087923292, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 1696.7760205006925, + "ttft_p95_ms": 7079.009777032479, + "ttft_p99_ms": 9906.125367596096, + "tpot_p50_ms": 18.908712023488928, + "tpot_p95_ms": 37.75210648097284, + "tpot_p99_ms": 42.41224298877765, + "e2e_p50_ms": 13078.812665509759, + "e2e_p95_ms": 24145.498285339272, + "e2e_p99_ms": 26969.445044811862, + "itl_p50_ms": 17.41874399886001, + "itl_p95_ms": 17.961374601873104, + "itl_p99_ms": 383.83775273512606, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o1024_a1.json new file mode 100644 index 0000000..01e65be --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 111.18827333900845, + "request_tps": 2.8780013430403155, + "input_tps": 2947.073375273283, + "output_tps": 2947.073375273283, + "total_tps": 5894.146750546566, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1236.4594759856118, + "ttft_p95_ms": 3486.339133232832, + "ttft_p99_ms": 3954.5976989151677, + "tpot_p50_ms": 20.184367791299923, + "tpot_p95_ms": 22.733724067629, + "tpot_p99_ms": 23.26845901167211, + "e2e_p50_ms": 21600.943245503004, + "e2e_p95_ms": 24639.851032342995, + "e2e_p99_ms": 24654.7415991014, + "itl_p50_ms": 19.554279977455735, + "itl_p95_ms": 20.101924979826435, + "itl_p99_ms": 21.4101209945511, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o2048_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o2048_a1.json new file mode 100644 index 0000000..1680863 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o2048_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 231.3560413280211, + "request_tps": 1.3831495307541928, + "input_tps": 1416.3451194922934, + "output_tps": 2832.690238984587, + "total_tps": 4249.03535847688, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 2048.0, + "ttft_p50_ms": 1237.1772454935126, + "ttft_p95_ms": 4075.03496848949, + "ttft_p99_ms": 4438.276356909774, + "tpot_p50_ms": 19.961317755979394, + "tpot_p95_ms": 29.797527954492573, + "tpot_p99_ms": 29.79867891234008, + "e2e_p50_ms": 41953.71764700394, + "e2e_p95_ms": 61770.79950050393, + "e2e_p99_ms": 61851.04672904883, + "itl_p50_ms": 19.631187984487042, + "itl_p95_ms": 20.16204949177336, + "itl_p99_ms": 20.984756899997592, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o4096_a1.json new file mode 100644 index 0000000..b2e0883 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c64_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 320, + "failed": 0, + "duration_s": 413.36956618897966, + "request_tps": 0.7741256884250303, + "input_tps": 792.704704947231, + "output_tps": 3170.818819788924, + "total_tps": 3963.523524736155, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 1353.9671769976849, + "ttft_p95_ms": 3372.6759848868824, + "ttft_p99_ms": 3923.934282742266, + "tpot_p50_ms": 19.719561329914566, + "tpot_p95_ms": 20.46792597871924, + "tpot_p99_ms": 20.599813236414754, + "e2e_p50_ms": 82000.917450001, + "e2e_p95_ms": 85134.73726001248, + "e2e_p99_ms": 85152.08560849074, + "itl_p50_ms": 19.557946507120505, + "itl_p95_ms": 20.046211070439313, + "itl_p99_ms": 20.78921643726062, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o1024_a1.json new file mode 100644 index 0000000..faa113b --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 400, + "failed": 0, + "duration_s": 131.46577305399114, + "request_tps": 3.0426170303332536, + "input_tps": 3115.6398390612517, + "output_tps": 3115.6398390612517, + "total_tps": 6231.279678122503, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 1146.1891874932917, + "ttft_p95_ms": 4487.726777615899, + "ttft_p99_ms": 5507.241578921676, + "tpot_p50_ms": 23.907261485326536, + "tpot_p95_ms": 26.730242256350078, + "tpot_p99_ms": 27.79943066497788, + "e2e_p50_ms": 25425.060303488863, + "e2e_p95_ms": 29093.009902854097, + "e2e_p99_ms": 29212.41997377714, + "itl_p50_ms": 23.131843001465313, + "itl_p95_ms": 27.57070749357808, + "itl_p99_ms": 28.485130463086534, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o4096_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o4096_a1.json new file mode 100644 index 0000000..0601b4d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c80_i1024_o4096_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 400, + "failed": 0, + "duration_s": 501.7031515800045, + "request_tps": 0.797284208281904, + "input_tps": 816.4190292806697, + "output_tps": 3265.676117122679, + "total_tps": 4082.095146403349, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 4096.0, + "ttft_p50_ms": 2250.2203854965046, + "ttft_p95_ms": 4658.4656137551065, + "ttft_p99_ms": 5716.611436419189, + "tpot_p50_ms": 24.059723582905082, + "tpot_p95_ms": 24.354225822548287, + "tpot_p99_ms": 24.487317885165243, + "e2e_p50_ms": 99781.74352001224, + "e2e_p95_ms": 103402.10737639281, + "e2e_p99_ms": 104522.76062945835, + "itl_p50_ms": 23.14190100878477, + "itl_p95_ms": 23.698028920625802, + "itl_p99_ms": 24.58483098977012, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o128_a1.json new file mode 100644 index 0000000..ce3ca72 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 73.70493139498285, + "request_tps": 0.5427045279459121, + "input_tps": 555.729436616614, + "output_tps": 69.46617957707674, + "total_tps": 625.1956161936906, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 3509.9659979896387, + "ttft_p95_ms": 27374.682503790245, + "ttft_p99_ms": 32961.56131610186, + "tpot_p50_ms": 9.04847947639087, + "tpot_p95_ms": 249.60066149328418, + "tpot_p99_ms": 278.0204192110943, + "e2e_p50_ms": 4659.724333003396, + "e2e_p95_ms": 37692.61776938074, + "e2e_p99_ms": 37693.20461160998, + "itl_p50_ms": 9.10900448798202, + "itl_p95_ms": 9.500342964020094, + "itl_p99_ms": 10.415470011648722, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o256_a1.json new file mode 100644 index 0000000..53ed074 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i1024_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 23.89860760301235, + "request_tps": 1.6737376781297548, + "input_tps": 1713.907382404869, + "output_tps": 428.47684560121724, + "total_tps": 2142.3842280060862, + "mean_input_tokens": 1024.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 240.56512699462473, + "ttft_p95_ms": 10569.026969805418, + "ttft_p99_ms": 10732.348576473596, + "tpot_p50_ms": 9.789249927430944, + "tpot_p95_ms": 10.445169046887552, + "tpot_p99_ms": 35.187820401407905, + "e2e_p50_ms": 2737.936751509551, + "e2e_p95_ms": 13064.727992002736, + "e2e_p99_ms": 13066.108254613355, + "itl_p50_ms": 9.119076014030725, + "itl_p95_ms": 9.404584451112896, + "itl_p99_ms": 10.967337382608093, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i131072_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i131072_o128_a1.json new file mode 100644 index 0000000..5bfb77d --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i131072_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 542.2442093600112, + "request_tps": 0.07376750052750287, + "input_tps": 9668.853829140857, + "output_tps": 9.442240067520368, + "total_tps": 9678.296069208376, + "mean_input_tokens": 131072.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 56039.164249494206, + "ttft_p95_ms": 72600.3135224935, + "ttft_p99_ms": 92221.81805258355, + "tpot_p50_ms": 387.13930614164747, + "tpot_p95_ms": 753.8479004563203, + "tpot_p99_ms": 754.9241562044031, + "e2e_p50_ms": 110909.34608949465, + "e2e_p95_ms": 112789.03825180455, + "e2e_p99_ms": 134558.94819398702, + "itl_p50_ms": 660.2215159800835, + "itl_p95_ms": 1072.356289398158, + "itl_p99_ms": 1163.3652878389696, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o1024_a1.json new file mode 100644 index 0000000..eae32b9 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 50.61696275198483, + "request_tps": 0.7902489170674606, + "input_tps": 12947.438257233274, + "output_tps": 809.2148910770796, + "total_tps": 13756.653148310354, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 573.4946330048842, + "ttft_p95_ms": 733.3654505957384, + "ttft_p99_ms": 790.053364574851, + "tpot_p50_ms": 9.346392439878139, + "tpot_p95_ms": 9.664219554327293, + "tpot_p99_ms": 9.85550387294379, + "e2e_p50_ms": 10074.975329000154, + "e2e_p95_ms": 10331.053190038074, + "e2e_p99_ms": 10332.716352115385, + "itl_p50_ms": 9.268547495594248, + "itl_p95_ms": 9.488147572847083, + "itl_p99_ms": 10.121553320495892, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o128_a1.json new file mode 100644 index 0000000..412718e --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 8.840474782977253, + "request_tps": 4.524643866076273, + "input_tps": 74131.76510179366, + "output_tps": 579.154414857763, + "total_tps": 74710.91951665143, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 611.7119685077341, + "ttft_p95_ms": 705.7421804507612, + "ttft_p99_ms": 728.9413817677995, + "tpot_p50_ms": 8.631740236223134, + "tpot_p95_ms": 11.94178506870173, + "tpot_p99_ms": 12.782024573981596, + "e2e_p50_ms": 1712.7858160092728, + "e2e_p95_ms": 1891.0789757341263, + "e2e_p99_ms": 1892.5092905788915, + "itl_p50_ms": 9.147118500550278, + "itl_p95_ms": 9.401795730809681, + "itl_p99_ms": 14.493186495383268, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o512_a1.json new file mode 100644 index 0000000..1615f5e --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i16384_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 73.81995394398109, + "request_tps": 0.5418589129756752, + "input_tps": 8877.816430193463, + "output_tps": 277.4317634435457, + "total_tps": 9155.248193637008, + "mean_input_tokens": 16384.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 5848.043638005038, + "ttft_p95_ms": 10270.304282526195, + "ttft_p99_ms": 10305.013074664457, + "tpot_p50_ms": 17.2701541448243, + "tpot_p95_ms": 26.505919335452827, + "tpot_p99_ms": 26.557917271059214, + "e2e_p50_ms": 14963.280752504943, + "e2e_p95_ms": 15029.205415995966, + "e2e_p99_ms": 15033.096053168701, + "itl_p50_ms": 9.254616481484845, + "itl_p95_ms": 9.525524696800858, + "itl_p99_ms": 610.016989769938, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i4096_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i4096_o128_a1.json new file mode 100644 index 0000000..546dac9 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i4096_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 8.121882901003119, + "request_tps": 4.924966351713797, + "input_tps": 20172.662176619713, + "output_tps": 630.395693019366, + "total_tps": 20803.05786963908, + "mean_input_tokens": 4096.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 391.37257999391295, + "ttft_p95_ms": 654.7860209611827, + "ttft_p99_ms": 655.5212389715598, + "tpot_p50_ms": 9.02552549213017, + "tpot_p95_ms": 11.834378843241568, + "tpot_p99_ms": 12.646475187780089, + "e2e_p50_ms": 1594.1091245040298, + "e2e_p95_ms": 1801.0780599128339, + "e2e_p99_ms": 1801.8622757561388, + "itl_p50_ms": 9.091041501960717, + "itl_p95_ms": 9.295591751288157, + "itl_p99_ms": 11.05267516104506, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o1024_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o1024_a1.json new file mode 100644 index 0000000..f9ce42b --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o1024_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 278.18836335701053, + "request_tps": 0.14378746658308766, + "input_tps": 9423.255409989233, + "output_tps": 147.23836578108177, + "total_tps": 9570.493775770316, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 1024.0, + "ttft_p50_ms": 24218.15920100198, + "ttft_p95_ms": 37847.32255409762, + "ttft_p99_ms": 40382.343416879594, + "tpot_p50_ms": 29.127336590921143, + "tpot_p95_ms": 49.198489598973694, + "tpot_p99_ms": 49.27219220356028, + "e2e_p50_ms": 56596.45105450181, + "e2e_p95_ms": 56909.13160444033, + "e2e_p99_ms": 62302.39748750086, + "itl_p50_ms": 9.62444901233539, + "itl_p95_ms": 9.987523194286041, + "itl_p99_ms": 776.7235341103515, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o128_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o128_a1.json new file mode 100644 index 0000000..d31da75 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o128_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 235.13304557901574, + "request_tps": 0.17011645428867683, + "input_tps": 11148.751948262725, + "output_tps": 21.774906148950635, + "total_tps": 11170.526854411675, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 128.0, + "ttft_p50_ms": 24172.63857051148, + "ttft_p95_ms": 35960.15721101285, + "ttft_p99_ms": 39592.695025814464, + "tpot_p50_ms": 166.83349701175814, + "tpot_p95_ms": 328.55576824153474, + "tpot_p99_ms": 329.2510806872486, + "e2e_p50_ms": 48098.485034002806, + "e2e_p95_ms": 51191.40079334465, + "e2e_p99_ms": 56645.06894614664, + "itl_p50_ms": 9.62428402272053, + "itl_p95_ms": 816.3079831079814, + "itl_p99_ms": 889.5722046121956, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o256_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o256_a1.json new file mode 100644 index 0000000..a57663f --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o256_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 241.68815920100315, + "request_tps": 0.1655025224745639, + "input_tps": 10846.37331289302, + "output_tps": 42.36864575348836, + "total_tps": 10888.741958646508, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 256.0, + "ttft_p50_ms": 24228.64825501165, + "ttft_p95_ms": 35330.018459535495, + "ttft_p99_ms": 39612.372868892846, + "tpot_p50_ms": 88.08975900584996, + "tpot_p95_ms": 168.72122898116345, + "tpot_p99_ms": 169.04595058763002, + "e2e_p50_ms": 49360.89754800196, + "e2e_p95_ms": 52881.204059695294, + "e2e_p99_ms": 57884.091986017, + "itl_p50_ms": 9.613202011678368, + "itl_p95_ms": 762.7214051928601, + "itl_p99_ms": 843.184829771053, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o512_a1.json b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o512_a1.json new file mode 100644 index 0000000..b06ce9a --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/metrics/vllm_c8_i65536_o512_a1.json @@ -0,0 +1,25 @@ +{ + "status": "COMPLETED", + "completed": 40, + "failed": 0, + "duration_s": 253.59146945900284, + "request_tps": 0.15773401244660815, + "input_tps": 10337.256239700911, + "output_tps": 80.75981437266337, + "total_tps": 10418.016054073574, + "mean_input_tokens": 65536.0, + "mean_output_tokens": 512.0, + "ttft_p50_ms": 18460.19749098923, + "ttft_p95_ms": 35239.79678586038, + "ttft_p99_ms": 39810.581101148564, + "tpot_p50_ms": 60.03944050394105, + "tpot_p95_ms": 88.83629755211601, + "tpot_p99_ms": 89.054011180707, + "e2e_p50_ms": 51667.23371298576, + "e2e_p95_ms": 64968.29380615088, + "e2e_p99_ms": 74357.24756326554, + "itl_p50_ms": 9.60926100378856, + "itl_p95_ms": 689.5956993947038, + "itl_p99_ms": 821.5999770630151, + "validation_errors": [] +} diff --git a/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/server_cmd.txt b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/server_cmd.txt new file mode 100644 index 0000000..3450120 --- /dev/null +++ b/experiments/h20/dsv4_h20_vllm_tp_dp_matrix/adaptive_results/adaptive_20260716-051910/tp8_dp1/server_cmd.txt @@ -0,0 +1 @@ +vllm serve /data1/hf_models/DeepSeek-V4-Flash --trust-remote-code --kv-cache-dtype fp8 --block-size 256 --tensor-parallel-size 8 --gpu-memory-utilization 0.9 --max-model-len 1048576 --max-num-seqs 128 --no-enable-flashinfer-autotune --host 0.0.0.0 --port 30030