NVFP4 dual GEMM
NVFP4 block-scaled dual matrix multiply with SiLU for B200: C = silu(A @ B1) * (A @ B2), with A (m×k×l), B1 and B2 (n×k×l) in nvfp4 e2m1, fp8 scale factors per 16 elements of k, and fp16 output (m×n×l). All operands are K-major; m and n are divisible by the kernel's mma tile and k by 256.
Reported evidence · last observed 2026-01-20. Reported by source; not independently reproduced. License unknown.
Current records
Not measured on H100 for this workload. Challenges →
1submission 384889guaguabear12.9µs1.00×Reported · license unknown · source2026-01-20stale
Measured exactly what you asked. Reported by source; not independently reproduced.
2submission 379817gau.nernst13.1µs1.02×Reported · license unknown · source2026-01-19stale
Measured exactly what you asked. Reported by source; not independently reproduced.
3submission 382690arseni_ivanov13.8µs1.07×Reported · license unknown · source2026-01-19stale
Measured exactly what you asked. Reported by source; not independently reproduced.
4submission 277807guaguabear13.8µs1.07×Reported · license unknown · source2026-01-05stale
Measured exactly what you asked. Reported by source; not independently reproduced.
5submission 291302Simon14.1µs1.09×Reported · license unknown · source2026-01-06stale
Measured exactly what you asked. Reported by source; not independently reproduced.
6submission 258061Simon14.1µs1.09×Reported · license unknown · source2026-01-02stale
Measured exactly what you asked. Reported by source; not independently reproduced.
7submission 380730macto14.1µs1.10×Reported · license unknown · source2026-01-19stale
Measured exactly what you asked. Reported by source; not independently reproduced.
8submission 247703Simon14.1µs1.10×Reported · license unknown · source2026-01-01stale
Measured exactly what you asked. Reported by source; not independently reproduced.
9submission 280760rex_cz14.2µs1.10×Reported · license unknown · source2026-01-05stale
Measured exactly what you asked. Reported by source; not independently reproduced.
10submission 316443jiab_8528114.2µs1.10×Reported · license unknown · source2026-01-09stale
Measured exactly what you asked. Reported by source; not independently reproduced.
11submission 297880Arseni Ivanov14.2µs1.10×Reported · license unknown · source2026-01-07stale
Measured exactly what you asked. Reported by source; not independently reproduced.
12submission 380607macto14.2µs1.10×Reported · license unknown · source2026-01-19stale
Measured exactly what you asked. Reported by source; not independently reproduced.
13submission 221602Simon14.2µs1.10×Reported · license unknown · source2025-12-27stale
Measured exactly what you asked. Reported by source; not independently reproduced.
14submission 249721Arseni Ivanov14.2µs1.10×Reported · license unknown · source2026-01-01stale
Measured exactly what you asked. Reported by source; not independently reproduced.
15submission 316390jiab_8528114.2µs1.10×Reported · license unknown · source2026-01-09stale
Measured exactly what you asked. Reported by source; not independently reproduced.
16submission 316261jiab_8528114.2µs1.10×Reported · license unknown · source2026-01-09stale
Measured exactly what you asked. Reported by source; not independently reproduced.
17submission 279897rex_cz14.2µs1.10×Reported · license unknown · source2026-01-05stale
Measured exactly what you asked. Reported by source; not independently reproduced.
18submission 383988crUst14.2µs1.10×Reported · license unknown · source2026-01-20stale
Measured exactly what you asked. Reported by source; not independently reproduced.
19submission 340117basesearch14.2µs1.10×Reported · license unknown · source2026-01-13stale
Measured exactly what you asked. Reported by source; not independently reproduced.
20submission 368718arseni_ivanov14.2µs1.10×Reported · license unknown · source2026-01-17stale
Measured exactly what you asked. Reported by source; not independently reproduced.
21submission 299074jiab_8528114.2µs1.10×Reported · license unknown · source2026-01-07stale
Measured exactly what you asked. Reported by source; not independently reproduced.
22submission 296470shigao14.2µs1.10×Reported · license unknown · source2026-01-07stale
Measured exactly what you asked. Reported by source; not independently reproduced.
23submission 276911gau.nernst14.2µs1.10×Reported · license unknown · source2026-01-05stale
Measured exactly what you asked. Reported by source; not independently reproduced.
24submission 344607shiyegao14.2µs1.10×Reported · license unknown · source2026-01-14stale
Measured exactly what you asked. Reported by source; not independently reproduced.
25submission 365816crUst14.2µs1.10×Reported · license unknown · source2026-01-17stale
Measured exactly what you asked. Reported by source; not independently reproduced.
26submission 264542rex_cz14.2µs1.10×Reported · license unknown · source2026-01-03stale
Measured exactly what you asked. Reported by source; not independently reproduced.
27submission 275665guaguabear14.2µs1.10×Reported · license unknown · source2026-01-05stale
Measured exactly what you asked. Reported by source; not independently reproduced.
28submission 271285gau.nernst14.2µs1.10×Reported · license unknown · source2026-01-04stale
Measured exactly what you asked. Reported by source; not independently reproduced.
29submission 242254guaguabear14.2µs1.10×Reported · license unknown · source2025-12-31stale
Measured exactly what you asked. Reported by source; not independently reproduced.
30submission 253604gau.nernst14.2µs1.10×Reported · license unknown · source2026-01-02stale
Measured exactly what you asked. Reported by source; not independently reproduced.
31submission 242314Arseni Ivanov14.2µs1.10×Reported · license unknown · source2025-12-31stale
Measured exactly what you asked. Reported by source; not independently reproduced.
32submission 358880crUst14.2µs1.10×Reported · license unknown · source2026-01-16stale
Measured exactly what you asked. Reported by source; not independently reproduced.
33submission 339941novo_force14.2µs1.10×Reported · license unknown · source2026-01-13stale
Measured exactly what you asked. Reported by source; not independently reproduced.
34submission 253572gau.nernst14.2µs1.10×Reported · license unknown · source2026-01-02stale
Measured exactly what you asked. Reported by source; not independently reproduced.
35submission 242083Arseni Ivanov14.3µs1.10×Reported · license unknown · source2025-12-31stale
Measured exactly what you asked. Reported by source; not independently reproduced.
36submission 296743jiab_8528114.3µs1.11×Reported · license unknown · source2026-01-07stale
Measured exactly what you asked. Reported by source; not independently reproduced.
37submission 235613guaguabear14.3µs1.11×Reported · license unknown · source2025-12-29stale
Measured exactly what you asked. Reported by source; not independently reproduced.
38submission 264512rex_cz14.3µs1.11×Reported · license unknown · source2026-01-03stale
Measured exactly what you asked. Reported by source; not independently reproduced.
39submission 221524Simon14.3µs1.11×Reported · license unknown · source2025-12-27stale
Measured exactly what you asked. Reported by source; not independently reproduced.
40submission 233437guaguabear14.3µs1.11×Reported · license unknown · source2025-12-29stale
Measured exactly what you asked. Reported by source; not independently reproduced.
41submission 224818gau.nernst14.3µs1.11×Reported · license unknown · source2025-12-28stale
Measured exactly what you asked. Reported by source; not independently reproduced.
42submission 221124Simon14.4µs1.11×Reported · license unknown · source2025-12-27stale
Measured exactly what you asked. Reported by source; not independently reproduced.
43submission 358026crUst14.4µs1.12×Reported · license unknown · source2026-01-16stale
Measured exactly what you asked. Reported by source; not independently reproduced.
44submission 355420macto14.4µs1.12×Reported · license unknown · source2026-01-15stale
Measured exactly what you asked. Reported by source; not independently reproduced.
45submission 293471jiab_8528114.5µs1.12×Reported · license unknown · source2026-01-07stale
Measured exactly what you asked. Reported by source; not independently reproduced.
46submission 224570gau.nernst14.5µs1.12×Reported · license unknown · source2025-12-28stale
Measured exactly what you asked. Reported by source; not independently reproduced.
47submission 209319gum 九尾狐14.5µs1.12×Reported · license unknown · source2025-12-25stale
Measured exactly what you asked. Reported by source; not independently reproduced.
48submission 218992gau.nernst14.5µs1.12×Reported · license unknown · source2025-12-27stale
Measured exactly what you asked. Reported by source; not independently reproduced.
49submission 233411guaguabear14.6µs1.13×Reported · license unknown · source2025-12-29stale
Measured exactly what you asked. Reported by source; not independently reproduced.
5014.6µs1.13×Reported · license unknown · source2026-01-09stale
Measured exactly what you asked. Reported by source; not independently reproduced.
51submission 284405jiab_8528114.6µs1.13×Reported · license unknown · source2026-01-06stale
Measured exactly what you asked. Reported by source; not independently reproduced.
52submission 246839rex_cz14.6µs1.13×Reported · license unknown · source2026-01-01stale
Measured exactly what you asked. Reported by source; not independently reproduced.
53submission 215135Simon14.6µs1.13×Reported · license unknown · source2025-12-26stale
Measured exactly what you asked. Reported by source; not independently reproduced.
54submission 246586rex_cz14.6µs1.13×Reported · license unknown · source2026-01-01stale
Measured exactly what you asked. Reported by source; not independently reproduced.
55submission 244581rex_cz14.6µs1.13×Reported · license unknown · source2025-12-31stale
Measured exactly what you asked. Reported by source; not independently reproduced.
56submission 244409rex_cz14.6µs1.13×Reported · license unknown · source2025-12-31stale
Measured exactly what you asked. Reported by source; not independently reproduced.
57submission 232395guaguabear14.6µs1.13×Reported · license unknown · source2025-12-29stale
Measured exactly what you asked. Reported by source; not independently reproduced.
58submission 218765gau.nernst14.6µs1.13×Reported · license unknown · source2025-12-27stale
Measured exactly what you asked. Reported by source; not independently reproduced.
59submission 364470leymore417214.6µs1.13×Reported · license unknown · source2026-01-17stale
Measured exactly what you asked. Reported by source; not independently reproduced.
60submission 244261rex_cz14.6µs1.13×Reported · license unknown · source2025-12-31stale
Measured exactly what you asked. Reported by source; not independently reproduced.
360 more rows in this comparison group. Open all in search →