sglang/ GB200 perf

dsr1-fp4-1k1k-mid-curve

concurrency 8,192 · 4 days ago

cron
passed
re-run
Commit

[cookbook] Laguna-M.1: add PD disaggregation section (#28737)

Jiminator·4 days ago
PR #28737 · [cookbook] Laguna-M.1: add PD disaggregation section
Run
GitHub Actions27858605416-1
Slurm job5326
GPUs48·prefill 16 / decode 32
ISL / OSL1024 / 1024

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

81,920

duration

689

max_concurrency

8,192

mean_e2el_ms

66,542ms

mean_itl_ms

1,202ms

mean_tpot_ms

24.53ms

mean_ttft_ms

43,964ms

median_e2el_ms

69,530ms

median_itl_ms

1,041ms

median_tpot_ms

22.11ms

median_ttft_ms

49,390ms

num_prompts

81,920

output_throughput

109,639tok/s

p99_e2el_ms

81,028ms

p99_itl_ms

3,374ms

p99_tpot_ms

42.04ms

p99_ttft_ms

56,639ms

peak_output_tokens_per_s

154,505s

request_throughput

119tok/s

std_e2el_ms

8,677ms

std_itl_ms

623ms

std_tpot_ms

7.02ms

std_ttft_ms

11,678ms

total_input_tokens

75,511,905

total_output_tokens

75,496,082

total_token_throughput

219,301tok/s