sglang/ GB200 perf

dsr1-fp4-1k1k-mid-curve

concurrency 8,192 · 19 days ago

cron
passed
re-run
Commit

docs: sync legacy docs/-only updates into docs_new (Mintlify) (#27308)

zijiexia·19 days ago
PR #27308 · docs: sync legacy docs/-only updates into docs_new (Mintlify)
Run
GitHub Actions26993287488-1
Slurm job5119
GPUs48·prefill 16 / decode 32
ISL / OSL1024 / 1024

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

81,920

duration

685

max_concurrency

8,192

mean_e2el_ms

66,264ms

mean_itl_ms

1,113ms

mean_tpot_ms

22.72ms

mean_ttft_ms

45,349ms

median_e2el_ms

69,162ms

median_itl_ms

947ms

median_tpot_ms

20.08ms

median_ttft_ms

51,733ms

num_prompts

81,920

output_throughput

110,277tok/s

p99_e2el_ms

83,288ms

p99_itl_ms

3,170ms

p99_tpot_ms

39.31ms

p99_ttft_ms

55,840ms

peak_output_tokens_per_s

160,030s

request_throughput

120tok/s

std_e2el_ms

8,215ms

std_itl_ms

628ms

std_tpot_ms

6.92ms

std_ttft_ms

11,057ms

total_input_tokens

75,511,905

total_output_tokens

75,496,082

total_token_throughput

220,576tok/s