dsr1-fp4-1k1k-mid-curve

concurrency 2,048 · 18 days ago

cron

passed

re-run

Commit

SHAbf66b7b

[diffusion] model: support Ideogram4 NVFP4 (#27379)

mickqian·18 days ago

PR #27379 · [diffusion] model: support Ideogram4 NVFP4

Run

GitHub Actions27051102318-1

Slurm job5123

GPUs48·prefill 16 / decode 32

ISL / OSL1024 / 1024

Logscron/27051102318-1/1k1k/dsr1-fp4-1k1k-mid-curve/2026-06-06/5123/sa-bench_isl_1024_osl_1024/

Metrics

28 captured

best_of

1.00

burstiness

1.00

completed

20,480

duration

201

max_concurrency

2,048

mean_e2el_ms

19,362ms

mean_itl_ms

941ms

mean_tpot_ms

19.20ms

mean_ttft_ms

1,670ms

median_e2el_ms

18,954ms

median_itl_ms

974ms

median_tpot_ms

19.59ms

median_ttft_ms

785ms

num_prompts

20,480

output_throughput

94,128tok/s

p99_e2el_ms

30,840ms

p99_itl_ms

1,555ms

p99_tpot_ms

20.58ms

p99_ttft_ms

12,532ms

peak_output_tokens_per_s

108,148s

request_throughput

102tok/s

std_e2el_ms

2,837ms

std_itl_ms

192ms

std_tpot_ms

1.32ms

std_ttft_ms

2,630ms

total_input_tokens

18,880,692

total_output_tokens

18,888,974

total_token_throughput

188,215tok/s