GQA paged prefill causal h32 kv4 d128 ps1
Batched Grouped Query Attention prefill with a paged KV cache. Causal mask is applied. Captured from Qwen3-30B-A3B during incremental prefill.
Reported evidence · last observed 2026-08-17. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 30 cases · mean latency60 cases
Not measured on H100 for this workload. Challenges →
15.31µs99% SOL1.00×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
25.43µs99% SOL1.02×Reported · license unknown · no source2026-08-16
Measured exactly what you asked. Reported by source; not independently reproduced.
35.58µs99% SOL1.05×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
45.64µs98% SOL1.06×Reported · license unknown · no source2026-08-14
Measured exactly what you asked. Reported by source; not independently reproduced.
55.72µs98% SOL1.08×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
66.13µs98% SOL1.15×Reported · license unknown · no source2026-08-10
Measured exactly what you asked. Reported by source; not independently reproduced.
76.14µs98% SOL1.15×Reported · license unknown · no source2026-07-28
Measured exactly what you asked. Reported by source; not independently reproduced.
86.26µs98% SOL1.18×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
96.28µs97% SOL1.18×Reported · license unknown · no source2026-07-24
Measured exactly what you asked. Reported by source; not independently reproduced.
106.33µs98% SOL1.19×Reported · license unknown · no source2026-08-10
Measured exactly what you asked. Reported by source; not independently reproduced.
116.96µs98% SOL1.31×Reported · license unknown · no source2026-08-09
Measured exactly what you asked. Reported by source; not independently reproduced.
128.10µs98% SOL1.52×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
139.07µs98% SOL1.71×Reported · license unknown · no source2026-07-25
Measured exactly what you asked. Reported by source; not independently reproduced.
1411.0µs98% SOL2.07×Reported · license unknown · no source2026-07-31
Measured exactly what you asked. Reported by source; not independently reproduced.
1511.4µs96% SOL2.14×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
1613.4µs98% SOL2.52×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
1713.5µs97% SOL2.55×Reported · license unknown · no source2026-08-04
Measured exactly what you asked. Reported by source; not independently reproduced.
1813.6µs97% SOL2.55×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
1914.1µs98% SOL2.65×Reported · license unknown · no source2026-07-20
Measured exactly what you asked. Reported by source; not independently reproduced.
2014.2µs97% SOL2.67×Reported · license unknown · no source2026-07-15
Measured exactly what you asked. Reported by source; not independently reproduced.
2114.7µs96% SOL2.77×Reported · license unknown · no source2026-07-31
Measured exactly what you asked. Reported by source; not independently reproduced.
2215.0µs96% SOL2.81×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
2315.0µs97% SOL2.82×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2415.2µs97% SOL2.87×Reported · license unknown · no source2026-07-20
Measured exactly what you asked. Reported by source; not independently reproduced.
2515.3µs97% SOL2.88×Reported · license unknown · no source2026-07-19
Measured exactly what you asked. Reported by source; not independently reproduced.
2616.1µs97% SOL3.03×Reported · license unknown · no source2026-07-17
Measured exactly what you asked. Reported by source; not independently reproduced.
2716.7µs95% SOL3.14×Reported · license unknown · no source2026-07-16
Measured exactly what you asked. Reported by source; not independently reproduced.
2817.0µs95% SOL3.20×Reported · license unknown · no source2026-07-28
Measured exactly what you asked. Reported by source; not independently reproduced.
2917.7µs95% SOL3.34×Reported · license unknown · no source2026-07-16
Measured exactly what you asked. Reported by source; not independently reproduced.
3018.4µs95% SOL3.46×Reported · license unknown · no source2026-07-16
Measured exactly what you asked. Reported by source; not independently reproduced.
3119.5µs95% SOL3.68×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
3220.0µs96% SOL3.77×Reported · license unknown · no source2026-08-02
Measured exactly what you asked. Reported by source; not independently reproduced.
3320.1µs95% SOL3.78×Reported · license unknown · no source2026-07-18
Measured exactly what you asked. Reported by source; not independently reproduced.
3420.1µs95% SOL3.78×Reported · license unknown · no source2026-07-15
Measured exactly what you asked. Reported by source; not independently reproduced.
3520.2µs96% SOL3.79×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
3620.9µs94% SOL3.93×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
3722.5µs96% SOL4.24×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
3835.8µs91% SOL6.74×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
3940.3µs91% SOL7.59×Reported · license unknown · no source2026-07-29
Measured exactly what you asked. Reported by source; not independently reproduced.
4045.1µs92% SOL8.49×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
Implementations
Show all 60 implementations ›Showing all 60 implementations ⌄
Semantics
Also indexed as gqa-paged-prefill-causal-h32-kv4-d128-ps1 · reviewed as the same computation; every definition's cohorts are shown here, ranked separately.