qwen3-14b · KernelIndex

0 of 14 operations have a usable best known on NVIDIA B200 · 14 without one

Best known on NVIDIA B200

Operation · workload

Best known

Record

Trust

Observed

gemm4 operations

GEMM n34816 k5120fp16 · [4, 5120]

flashinfer / wrapper8028benot usable

72.4µs

Reported · Apache-2.0 · source

2026-03-23stale

No entry in this cohort passes the usability policy (no install recipe); the fastest known is shown. Measured by the source's own harness. Ranked by its metric under ranking-v1. 2 more entries in this cohort.

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

GEMM n5120 k17408fp16 · [4, 17408]

flashinfer / wrapperdd130anot usable

41.2µs

Reported · Apache-2.0 · source

2026-03-24stale

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

GEMM n5120 k5120fp16 · [7, 5120]

flashinfer / wrapperad9a00not usable

14.5µs

Reported · Apache-2.0 · source

2026-03-24stale

No entry in this cohort passes the usability policy (no install recipe); the fastest known is shown. Measured by the source's own harness. Ranked by its metric under ranking-v1. 1 more entry in this cohort.

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

GEMM n7168 k5120fp16 · [7, 5120]

flashinfer / wrapper4c2606not usable

18.1µs

Reported · Apache-2.0 · source

2026-03-24stale

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

gqa-paged-attention4 operations

GQA paged decode h20 kv4 d128 ps1bf16 · [1, 20, 128] · num pages 59 · num kv indices 58

flashinfer / wrapper94b73anot usable

25.6µs

Reported · Apache-2.0 · source

2026-03-28stale

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

GQA paged decode h20 kv4 d128 ps64bf16 · [1, 20, 128] · num pages 52 · num kv indices 51

flashinfer_wrapper_bcbabfnot usable

95.2µs

Reported · Apache-2.0 · source

2026-03-28stale

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

GQA paged prefill causal h20 kv4 d128 ps1bf16 · [1, 20, 128]

flashinfer / wrapper2d6239not usable

25.5µs

Reported · Apache-2.0 · source

2026-03-28stale

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

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

GQA paged prefill causal h20 kv4 d128 ps64bf16 · [1, 20, 128]

flashinfer / wrapperbe0f02not usable

24.3µs

Reported · Apache-2.0 · source

2026-03-28stale

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

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

gqa-ragged-attention1 operation

GQA ragged prefill causal h20 kv4 d128bf16 · [37, 4, 128]

flashinfer_wrapper_023122not usable

84.2µs

Reported · Apache-2.0 · source

2026-03-24stale

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

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →

rmsnorm1 operation

Fused add RMSNorm h5120bf16 · [5120] · batch 32

flashinfer / wrapper7f6051not usable

3.30µs

Reported · Apache-2.0 · source

2026-03-26stale

source mirroredApache-2.0no install recipeCohort on the operation page →View source →Vendor the source →Run detail →