Search · KernelIndex

Has source0License known0Installable038 hidden

Close matches. Each row lists what differs.

No compatible results for this workload under the active filters. 6 other protocols results → Best without source: 387.8 µs. Include results without source →

MLA paged decode h16 ckv512 kpe64 ps1Operation in the mla-paged-attention familyMLA fused QKV RoPE splitOperation in the mla-paged-attention familyMLA paged prefill causal h16 ckv512 kpe64 ps1Operation in the mla-paged-attention familyFP8 MLA KV compression projectionOperation in the mla-paged-attention familyMLA paged decode h16 ckv512 kpe64 ps1Operation in the mla-paged-attention familyMLA paged decode h16 ckv512 kpe64 ps64Operation in the mla-paged-attention family

Missing a kernel here? Evidence submissions open with the contribution beta.

This search as an API call

curlkicomparison key sha256:7b44b02df2216cf4…JSON

NVIDIA SOL-ExecBench · last observed 2026-08-24

Has source0License known0Installable038 hidden

Close matches. Each row lists what differs.

No compatible results for this workload under the active filters. 6 other protocols results → Best without source: 387.8 µs. Include results without source →

Missing a kernel here? Evidence submissions open with the contribution beta.

This search as an API call

curlkicomparison key sha256:7b44b02df2216cf4…JSON

NVIDIA SOL-ExecBench · last observed 2026-08-24

FP8 MLA attention output projection