FP8 attention output projection
FP8-quantized attention output projection that combines multi-head attention outputs back to hidden dimension. Projects concatenated attention outputs from all heads (num_heads * head_dim = 7680) to hidden_size (7680) using blockwise FP8 quantization with BlockWise1x128 for activations and BlockWise128x128 for weights.
Reported evidence · last observed 2026-08-23. Reported by source; not independently reproduced. No public source. License unknown.
Current records
Workloadsuite of 16 cases · mean latencyscale_n = 60 · head_dim = 128 · input_dim = 7680 · hidden_size = 7680 · scale_k_act = 60 · scale_k_weight = 60 · num_attention_heads = 6016 cases
Not measured on H100 for this workload. Challenges →
1252.0µs67% SOL1.00×Reported · license unknown · no source2026-08-23
Measured exactly what you asked. Reported by source; not independently reproduced.
2259.2µs67% SOL1.03×Reported · license unknown · no source2026-08-19
Measured exactly what you asked. Reported by source; not independently reproduced.
3264.2µs66% SOL1.05×Reported · license unknown · no source2026-07-31
Measured exactly what you asked. Reported by source; not independently reproduced.
4265.4µs65% SOL1.05×Reported · license unknown · no source2026-08-23
Measured exactly what you asked. Reported by source; not independently reproduced.
5276.9µs64% SOL1.10×Reported · license unknown · no source2026-07-29
Measured exactly what you asked. Reported by source; not independently reproduced.
6280.5µs64% SOL1.11×Reported · license unknown · no source2026-07-14
Measured exactly what you asked. Reported by source; not independently reproduced.
7292.2µs62% SOL1.16×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
8294.0µs62% SOL1.17×Reported · license unknown · no source2026-07-29
Measured exactly what you asked. Reported by source; not independently reproduced.
9309.6µs60% SOL1.23×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
10310.8µs60% SOL1.23×Reported · license unknown · no source2026-08-21
Measured exactly what you asked. Reported by source; not independently reproduced.
11312.1µs60% SOL1.24×Reported · license unknown · no source2026-08-22
Measured exactly what you asked. Reported by source; not independently reproduced.
12315.2µs59% SOL1.25×Reported · license unknown · no source2026-08-17
Measured exactly what you asked. Reported by source; not independently reproduced.
13325.0µs58% SOL1.29×Reported · license unknown · no source2026-08-22
Measured exactly what you asked. Reported by source; not independently reproduced.
14325.8µs58% SOL1.29×Reported · license unknown · no source2026-08-20
Measured exactly what you asked. Reported by source; not independently reproduced.
15326.1µs58% SOL1.29×Reported · license unknown · no source2026-08-23
Measured exactly what you asked. Reported by source; not independently reproduced.
16326.5µs58% SOL1.30×Reported · license unknown · no source2026-08-23
Measured exactly what you asked. Reported by source; not independently reproduced.
17326.5µs58% SOL1.30×Reported · license unknown · no source2026-08-22
Measured exactly what you asked. Reported by source; not independently reproduced.
18326.7µs58% SOL1.30×Reported · license unknown · no source2026-08-22
Measured exactly what you asked. Reported by source; not independently reproduced.
19327.2µs58% SOL1.30×Reported · license unknown · no source2026-08-11
Measured exactly what you asked. Reported by source; not independently reproduced.
20327.5µs58% SOL1.30×Reported · license unknown · no source2026-08-20
Measured exactly what you asked. Reported by source; not independently reproduced.
21328.2µs58% SOL1.30×Reported · license unknown · no source2026-08-11
Measured exactly what you asked. Reported by source; not independently reproduced.
22329.0µs58% SOL1.31×Reported · license unknown · no source2026-08-19
Measured exactly what you asked. Reported by source; not independently reproduced.
23329.5µs58% SOL1.31×Reported · license unknown · no source2026-08-22
Measured exactly what you asked. Reported by source; not independently reproduced.
24334.9µs57% SOL1.33×Reported · license unknown · no source2026-08-09
Measured exactly what you asked. Reported by source; not independently reproduced.
25347.7µs56% SOL1.38×Reported · license unknown · no source2026-08-24
Measured exactly what you asked. Reported by source; not independently reproduced.
26354.6µs56% SOL1.41×Reported · license unknown · no source2026-08-04
Measured exactly what you asked. Reported by source; not independently reproduced.
27361.1µs55% SOL1.43×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
28367.5µs54% SOL1.46×Reported · license unknown · no source2026-08-09
Measured exactly what you asked. Reported by source; not independently reproduced.
29395.6µs52% SOL1.57×Reported · license unknown · no source2026-08-15
Measured exactly what you asked. Reported by source; not independently reproduced.
30414.8µs50% SOL1.65×Reported · license unknown · no source2026-07-19
Measured exactly what you asked. Reported by source; not independently reproduced.
31439.2µs48% SOL1.74×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
32450.2µs48% SOL1.79×Reported · license unknown · no source2026-07-29
Measured exactly what you asked. Reported by source; not independently reproduced.
33466.6µs47% SOL1.85×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
34487.2µs45% SOL1.93×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
35490.2µs45% SOL1.95×Reported · license unknown · no source2026-08-02
Measured exactly what you asked. Reported by source; not independently reproduced.
36503.0µs44% SOL2.00×Reported · license unknown · no source2026-08-03
Measured exactly what you asked. Reported by source; not independently reproduced.
37626.9µs38% SOL2.49×Reported · license unknown · no source2026-07-26
Measured exactly what you asked. Reported by source; not independently reproduced.
381.04ms26% SOL4.14×Reported · license unknown · no source2026-07-26
Measured exactly what you asked. Reported by source; not independently reproduced.
391.05ms26% SOL4.18×Reported · license unknown · no source2026-07-23
Measured exactly what you asked. Reported by source; not independently reproduced.
402.05ms15% SOL8.14×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.
412.53ms12% SOL10.04×Reported · license unknown · no source2026-07-10
Measured exactly what you asked. Reported by source; not independently reproduced.