qwen3-vl-30b-a3b-instruct-fp8
0 of 7 operations have a usable best known on NVIDIA B200 · 7 without one
Best known on NVIDIA B200
FP8 attention output projectionsuite of 16 casesAmir M. Mir | SF Tensornot usable252.0µsReported · license unknown · no source2026-08-23
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. 40 more entries in this cohort.
FP8 attention QKV projectionsuite of 16 casesRIAC as wellnot usable233.2µsReported · license unknown · no source2026-08-19
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. 36 more entries in this cohort.
FP8 MoE expert linearsuite of 16 casesLasernot usable93.6µsReported · license unknown · no source2026-08-23
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. 21 more entries in this cohort.
FP8 MoE router projectionsuite of 16 casesRIAC as wellnot usable8.69µsReported · license unknown · no source2026-08-22
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. 20 more entries in this cohort.
FP8 MLP gate up projectionsuite of 16 casesRIAC as wellnot usable288.8µsReported · license unknown · no source2026-08-21
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. 43 more entries in this cohort.
FP8 vision attention output projectionsuite of 16 casesAmir M. Mir | SF Tensornot usable17.3µsReported · license unknown · no source2026-08-20
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.
FP8 vision attention QKV projectionsuite of 16 casesRIAC as wellnot usable29.5µsReported · license unknown · no source2026-08-19
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. 16 more entries in this cohort.
Coverage gaps on NVIDIA B200
No eligible evidence yet for these operations. Challenges →