History

yy-fighting 1482601ae5 Add num_prompts = 5×concurrency convention and update all experiments

- Document the rule in docs/EXPERIMENT_GUIDE.md section 7
- Update all config.env SCENARIOS to follow the 5x rule
- Migrate dsv4_h200_sglang, dsv4_h200_vllm, dsv4_p800_sglang
  from global NUM_PROMPTS to per-scenario num_prompts
- Keep long-context phase2 as exception (low counts documented)
- Add legacy 3-field fallback in run_bench.sh loops

2026-07-09 09:26:41 +00:00

config.env

Add num_prompts = 5×concurrency convention and update all experiments

2026-07-09 09:26:41 +00:00

parse_results.py

refactor: experiments/ + platforms/ layout and P800 SGLang support

2026-07-08 05:15:59 +00:00

README.md

fix(p800): managed-mode server startup and smoke-test defaults

2026-07-08 05:49:55 +00:00

run_bench.sh

Add num_prompts = 5×concurrency convention and update all experiments

2026-07-09 09:26:41 +00:00

start_server.sh

fix(p800): define RUN_ID/RESULT_ROOT/LOG_DIR in start_server.sh

2026-07-09 02:53:38 +00:00

README.md

DSV4 P800 SGLang Benchmark

Kunlun P800 XPU + SGLang + DeepSeek-V4-Flash-INT8 benchmark experiment.

Quick Start

# Run the default smoke-test scenario (starts server, runs benchmark, stops server)
PLATFORM=kunlun_p800 bash experiments/dsv4_p800_sglang/run_bench.sh

# Reuse an already-running server
SKIP_MANAGE_SERVER=1 PLATFORM=kunlun_p800 bash experiments/dsv4_p800_sglang/run_bench.sh

Results land in experiments/dsv4_p800_sglang/results/<RUN_ID>/.

Configuration

Edit config.env or override via environment variables:

# Run a full throughput grid instead of the default single smoke-test scenario
SCENARIOS=("32 512 256" "128 512 256" "256 512 256" "512 512 256") \
NUM_PROMPTS=512 \
PLATFORM=kunlun_p800 \
bash experiments/dsv4_p800_sglang/run_bench.sh

Files

File	Purpose
`config.env`	Experiment-level configuration (model, port, scenarios)
`start_server.sh`	Start the P800 SGLang Docker container
`run_bench.sh`	Orchestrator: server → benchmark → stop server
`parse_results.py`	Parse logs and generate `results.json` + `report.md`

Platform

This experiment targets platforms/kunlun_p800.env.