sglang/ GB200 perf

dsr1-fp4-1k1k-mid-curve

concurrency 512 · 14 hours ago

cron
passed
re-run
Commit

[AMD] Register DP attention test (#28495)

bingxche·16 hours ago
PR #28495 · [AMD] Register DP attention test
Run
GitHub Actions27927465630-1
Slurm job5332
GPUs48·prefill 16 / decode 32
ISL / OSL1024 / 1024

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

5,120

duration

164

max_concurrency

512

mean_e2el_ms

15,725ms

mean_itl_ms

783ms

mean_tpot_ms

15.97ms

mean_ttft_ms

1,002ms

median_e2el_ms

15,675ms

median_itl_ms

776ms

median_tpot_ms

16.21ms

median_ttft_ms

766ms

num_prompts

5,120

output_throughput

28,872tok/s

p99_e2el_ms

18,411ms

p99_itl_ms

1,198ms

p99_tpot_ms

16.67ms

p99_ttft_ms

3,128ms

peak_output_tokens_per_s

35,715s

request_throughput

31.30tok/s

std_e2el_ms

1,163ms

std_itl_ms

167ms

std_tpot_ms

0.79ms

std_ttft_ms

856ms

total_input_tokens

4,717,859

total_output_tokens

4,722,209

total_token_throughput

57,716tok/s