GQA paged decode h5 kv1 d128 ps1
20 eligible runs
gqa-paged-attention
Batched Grouped Query Attention decode with a paged KV cache. Captured from Llama 4 Scout 17B-16E with tensor parallel size 8 (40/8=5 q-heads, 8/8=1 kv-head).
Source baseline · unbeaten
83.4µsmean
flashinfer / wrapper8d9ac8FlashInfer-Bench baselines · Apache-2.0 · python
Reported evidence · last observed 2026-04-12. The source's designated baseline implementation. Reported by source; not independently reproduced.
Current records
Workloadnum_pages = 96 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 91 · bf1620 cases
num_pages
num_kv_indices
batch_size
len_indptr
dtype
scrolls · 20 cases total
Not measured on H100, B200 for this workload. Challenges →
Source-native comparison · GPU NVIDIA GB200 · Workload num_pages = 96 · batch_size = 1 · len_indptr = 2 · num_kv_indices = 91 · bf16 · CUDA 13.0 · Framework pytorch 2.9.1+cu130 · Protocol flashinfer-bench · mean · 1 results · last observed 2026-04-12Record history →
Estimated floor 6 ns · record 13136.6× above itestimate, not evidence ›
DRAM 6 ns · bandwidth-bound on GB200
every declared tensor crosses HBM exactly once (8,000 GB/s, GB200 datasheet)
no arithmetic formula for this family: bandwidth floor only
headroom-v1: a lower bound from declared tensors and datasheet peaks. A kernel can sit well above it for good reasons.
#
Implementation
Latency
vs #1
Trust
Observed
183.4µs1.00×Reported · Apache-2.0 · source2026-04-12
Measured exactly what you asked. The source's designated baseline implementation. Reported by source; not independently reproduced.
source mirroredApache-2.0no install recipeView source →Run detail →
Implementations
Implementation
Runtime
Best latency
Evidence
Availability
flashinfer / wrapper8d9ac8FlashInfer-Bench baselines
python
83.4µs
1.00×
Reproduction-ready
Apache-2.0 · source
Semantics
Inputs and outputs
qbf16 [batch_size, num_qo_heads, head_dim]
k_cachebf16 [num_pages, page_size, num_kv_heads, head_dim]
v_cachebf16 [num_pages, page_size, num_kv_heads, head_dim]
kv_indptrint32 [len_indptr]
kv_indicesint32 [num_kv_indices]
sm_scalefp32 scalar
outputbf16 [batch_size, num_qo_heads, head_dim]
lsefp32 [batch_size, num_qo_heads]
Axes and behavior
head_dimconstant = 128
num_pagesvariable
page_sizeconstant = 1
batch_sizevariable
len_indptrvariable
num_kv_headsconstant = 1
num_qo_headsconstant = 5
num_kv_indicesvariable
determinismunspecified
constraintsNo mutation or aliasing
Identity
Sources: FlashInfer-Bench (2026-04-12) · Apache-2.0last observed 2026-04-12How records are decidedJSON