Skip to content
KernelIndex
Search⌘K
Sign in

nvidia-nemotron-3-nano-30b-a3b-nvfp4

best known per operation on the selected GPU
4 operations · 2 families · 112 eligible runs
GPUNVIDIA B200112

0 of 4 operations have a usable best known on NVIDIA B200 · 4 without one

Best known on NVIDIA B200

Operation · workload
Best known
Record
Trust
Observed
quantization3 operations
NVFP4 attention output projectionsuite of 16 cases
Amir M. Mir | SF Tensornot usable
62.2µs
Reported · license unknown · no source
2026-07-21

No entry in this cohort passes the usability policy (no public source, no install recipe, license unknown); the fastest known is shown. Measured by the source's own harness. Ranked by its metric under ranking-v1. 13 more entries in this cohort.

no sourcelicense unknownno install recipeCohort on the operation page →Run detail →
NVFP4 attention QKV projectionsuite of 16 cases
Amir M. Mir | SF Tensornot usable
47.3µs
Reported · license unknown · no source
2026-07-17

No entry in this cohort passes the usability policy (no public source, no install recipe, license unknown); the fastest known is shown. Measured by the source's own harness. Ranked by its metric under ranking-v1. 15 more entries in this cohort.

no sourcelicense unknownno install recipeCohort on the operation page →Run detail →
NVFP4 Mamba2 out projectionsuite of 16 cases
Amir M. Mir | SF Tensornot usable
13.9µs
Reported · license unknown · no source
2026-08-16

No entry in this cohort passes the usability policy (no public source, no install recipe, license unknown); the fastest known is shown. Measured by the source's own harness. Ranked by its metric under ranking-v1. 32 more entries in this cohort.

no sourcelicense unknownno install recipeCohort on the operation page →Run detail →
moe1 operation
NVFP4 MoE router projectionsuite of 16 cases
Hollow Jellyfishnot usable
22.7µs
Reported · license unknown · no source
2026-08-22

No entry in this cohort passes the usability policy (no public source, no install recipe, license unknown); the fastest known is shown. Measured by the source's own harness. Ranked by its metric under ranking-v1. 26 more entries in this cohort.

no sourcelicense unknownno install recipeCohort on the operation page →Run detail →

Coverage gaps on NVIDIA B200

Operation
Family
Status
NVFP4 attention output projection
quantization
measured · no usable entry (no public source, no install recipe, license unknown)
NVFP4 attention QKV projection
quantization
measured · no usable entry (no public source, no install recipe, license unknown)
NVFP4 Mamba2 out projection
quantization
measured · no usable entry (no public source, no install recipe, license unknown)
NVFP4 MoE router projection
moe
measured · no usable entry (no public source, no install recipe, license unknown)

No eligible evidence yet for these operations. Challenges →

Sources: NVIDIA SOL-ExecBench (2026-08-22)How comparison worksJSON