sglang/ GB200 perf

dsr1-fp4-1k1k-mid-curve

concurrency 4,096 · 3 days ago

cron
passed
re-run
Commit

[diffusion] fix: validate openai sampling dimensions (#28791)

mickqian·3 days ago
PR #28791 · Validate OpenAI sampling dimensions
Run
GitHub Actions27892080060-1
Slurm job5327
GPUs48·prefill 16 / decode 32
ISL / OSL1024 / 1024

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

40,960

duration

320

max_concurrency

4,096

mean_e2el_ms

31,129ms

mean_itl_ms

1,173ms

mean_tpot_ms

23.93ms

mean_ttft_ms

9,099ms

median_e2el_ms

30,959ms

median_itl_ms

1,162ms

median_tpot_ms

24.13ms

median_ttft_ms

9,611ms

num_prompts

40,960

output_throughput

117,990tok/s

p99_e2el_ms

48,891ms

p99_itl_ms

2,343ms

p99_tpot_ms

28.75ms

p99_ttft_ms

23,752ms

peak_output_tokens_per_s

152,378s

request_throughput

128tok/s

std_e2el_ms

4,304ms

std_itl_ms

349ms

std_tpot_ms

3.09ms

std_ttft_ms

4,050ms

total_input_tokens

37,769,666

total_output_tokens

37,742,239

total_token_throughput

236,066tok/s