sglang/ GB200 perf

dsr1-fp4-1k1k-mid-curve

concurrency 4,096 · 20 days ago

cron
passed
re-run
Commit

[CPU] upgrade dependent torch ver to PT2.12 (#21456)

ZailiWang·20 days ago
PR #21456 · [CPU] upgrade dependent torch ver to PT2.12
Run
GitHub Actions26928315759-1
Slurm job5116
GPUs48·prefill 16 / decode 32
ISL / OSL1024 / 1024

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

40,960

duration

290

max_concurrency

4,096

mean_e2el_ms

28,333ms

mean_itl_ms

1,112ms

mean_tpot_ms

22.70ms

mean_ttft_ms

7,438ms

median_e2el_ms

27,996ms

median_itl_ms

1,107ms

median_tpot_ms

22.99ms

median_ttft_ms

7,241ms

num_prompts

40,960

output_throughput

129,980tok/s

p99_e2el_ms

47,554ms

p99_itl_ms

2,171ms

p99_tpot_ms

27.06ms

p99_ttft_ms

23,302ms

peak_output_tokens_per_s

175,193s

request_throughput

141tok/s

std_e2el_ms

4,413ms

std_itl_ms

320ms

std_tpot_ms

3.10ms

std_ttft_ms

4,130ms

total_input_tokens

37,769,666

total_output_tokens

37,742,239

total_token_throughput

260,055tok/s