GQA paged decode h32 kv8 d128 ps1
Batched Grouped Query Attention decode with a paged KV cache. Captured from Llama-3.1-8B.
Reported evidence · last observed 2026-08-21. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 48 cases · mean latency96 cases
Not measured on H100 for this workload. Challenges →
114.0µs91% SOL1.00×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
214.2µs91% SOL1.01×Reported · license unknown · no source2026-08-13
Measured exactly what you asked. Reported by source; not independently reproduced.
314.2µs91% SOL1.01×Reported · license unknown · no source2026-08-13
Measured exactly what you asked. Reported by source; not independently reproduced.
414.3µs91% SOL1.02×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
515.8µs90% SOL1.12×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
615.9µs90% SOL1.13×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
715.9µs90% SOL1.13×Reported · license unknown · no source2026-08-02
Measured exactly what you asked. Reported by source; not independently reproduced.
816.4µs90% SOL1.17×Reported · license unknown · no source2026-08-01
Measured exactly what you asked. Reported by source; not independently reproduced.
916.4µs90% SOL1.17×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
1016.4µs90% SOL1.17×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
1116.8µs89% SOL1.20×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
1217.1µs89% SOL1.22×Reported · license unknown · no source2026-07-30
Measured exactly what you asked. Reported by source; not independently reproduced.
1317.3µs89% SOL1.23×Reported · license unknown · no source2026-08-05
Measured exactly what you asked. Reported by source; not independently reproduced.
1418.1µs88% SOL1.29×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
1518.1µs89% SOL1.29×Reported · license unknown · no source2026-07-21
Measured exactly what you asked. Reported by source; not independently reproduced.
1618.2µs89% SOL1.29×Reported · license unknown · no source2026-07-21
Measured exactly what you asked. Reported by source; not independently reproduced.
1718.3µs88% SOL1.30×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
1818.6µs88% SOL1.32×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
1919.2µs87% SOL1.37×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2020.5µs87% SOL1.46×Reported · license unknown · no source2026-07-18
Measured exactly what you asked. Reported by source; not independently reproduced.
2120.6µs87% SOL1.47×Reported · license unknown · no source2026-07-16
Measured exactly what you asked. Reported by source; not independently reproduced.
2222.1µs85% SOL1.57×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
2323.3µs85% SOL1.66×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2427.2µs82% SOL1.94×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
2536.3µs77% SOL2.59×Reported · license unknown · no source2026-07-21
Measured exactly what you asked. Reported by source; not independently reproduced.
2639.1µs75% SOL2.78×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
27102.6µs54% SOL7.31×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
Implementations
Show all 53 implementations ›Showing all 53 implementations ⌄
Semantics
Also indexed as 013-gqa-paged-decode-h32-kv8-d128-ps1 · reviewed as the same computation; every definition's cohorts are shown here, ranked separately.