MLA paged prefill causal h16 ckv512 kpe64 ps1
Batched Multi-head Latent Attention prefill with a paged KV cache. Causal mask is applied. Captured from DeepSeek-V3 during incremental prefill with tensor parallel size 8.
Reported evidence · last observed 2026-08-16. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 38 cases · mean latencynum_pages = 98966976 cases
Not measured on H100 for this workload. Challenges →
131.1µs98% SOL1.00×Reported · license unknown · no source2026-08-16
Measured exactly what you asked. Reported by source; not independently reproduced.
231.3µs98% SOL1.01×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
332.0µs98% SOL1.03×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
436.0µs98% SOL1.16×Reported · license unknown · no source2026-07-29
Measured exactly what you asked. Reported by source; not independently reproduced.
536.1µs98% SOL1.16×Reported · license unknown · no source2026-07-31
Measured exactly what you asked. Reported by source; not independently reproduced.
637.5µs97% SOL1.20×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
739.4µs97% SOL1.27×Reported · license unknown · no source2026-08-02
Measured exactly what you asked. Reported by source; not independently reproduced.
839.5µs97% SOL1.27×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
939.6µs97% SOL1.27×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
1041.5µs97% SOL1.33×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
1141.8µs97% SOL1.34×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
1242.1µs97% SOL1.35×Reported · license unknown · no source2026-08-02
Measured exactly what you asked. Reported by source; not independently reproduced.
1343.2µs97% SOL1.39×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
1444.4µs97% SOL1.43×Reported · license unknown · no source2026-08-07
Measured exactly what you asked. Reported by source; not independently reproduced.
1544.7µs98% SOL1.44×Reported · license unknown · no source2026-07-24
Measured exactly what you asked. Reported by source; not independently reproduced.
1645.9µs97% SOL1.48×Reported · license unknown · no source2026-07-18
Measured exactly what you asked. Reported by source; not independently reproduced.
1747.3µs96% SOL1.52×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
1847.6µs96% SOL1.53×Reported · license unknown · no source2026-07-19
Measured exactly what you asked. Reported by source; not independently reproduced.
1949.1µs96% SOL1.58×Reported · license unknown · no source2026-07-29
Measured exactly what you asked. Reported by source; not independently reproduced.
2052.0µs97% SOL1.67×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2152.6µs96% SOL1.69×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2254.0µs97% SOL1.74×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
2354.6µs97% SOL1.75×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
2455.1µs96% SOL1.77×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2575.3µs95% SOL2.42×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
2693.3µs84% SOL3.00×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
27105.3µs93% SOL3.38×Reported · license unknown · no source2026-07-21
Measured exactly what you asked. Reported by source; not independently reproduced.
28147.8µs83% SOL4.75×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
29159.5µs90% SOL5.13×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
30263.6µs79% SOL8.47×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
Implementations
Show all 68 implementations ›Showing all 68 implementations ⌄
Semantics
Also indexed as mla-paged-prefill-causal-h16-ckv512-kpe64-ps1 · reviewed as the same computation; every definition's cohorts are shown here, ranked separately.