PyTorch
python · BSD-3-Clause
holds 21 records
Use it
Source available upstream · BSD-3-ClauseView source →
No install recipe recorded for this revision.
interfacepython
Compatibility
measured onNVIDIA H100, NVIDIA GeForce RTX 3090
declared hardwaredeclared only
architectures—
dtypes
Benchmark evidence
168 measurements across 2 GPUs, fastest first.
Operation / workload
Hardware
Latency
Rank
Observed
Show all 168 measurements ›Showing all 168 measurements ⌄
Fused neighborhood attentionfp32 · [512] · seq=256 · dilation=1
NVIDIA GeForce RTX 3090
547.0µs
#1 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=128 · dilation=1
NVIDIA GeForce RTX 3090
555.0µs
#1 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=256 · dilation=2
NVIDIA GeForce RTX 3090
556.1µs
#1 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=128 · dilation=2
NVIDIA GeForce RTX 3090
556.6µs
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=64 · dilation=2
NVIDIA GeForce RTX 3090
597.0µs
#1 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=64 · dilation=1
NVIDIA GeForce RTX 3090
600.1µs
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=512 · dilation=1
NVIDIA GeForce RTX 3090
871.4µs
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=512 · dilation=2
NVIDIA GeForce RTX 3090
871.4µs
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=1024 · dilation=2
NVIDIA GeForce RTX 3090
2.37ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=1024 · dilation=1
NVIDIA GeForce RTX 3090
2.37ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=64 · dilation=2
NVIDIA GeForce RTX 3090
5.06ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=64 · dilation=1
NVIDIA GeForce RTX 3090
5.17ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=64 · dilation=2
NVIDIA GeForce RTX 3090
6.21ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=64 · dilation=1
NVIDIA GeForce RTX 3090
6.84ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=2048 · dilation=1
NVIDIA GeForce RTX 3090
8.02ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=2048 · dilation=2
NVIDIA GeForce RTX 3090
8.03ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=128 · dilation=1
NVIDIA GeForce RTX 3090
10.2ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=128 · dilation=2
NVIDIA GeForce RTX 3090
10.4ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=128 · dilation=2
NVIDIA GeForce RTX 3090
11.4ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=128 · dilation=1
NVIDIA GeForce RTX 3090
11.8ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=256 · dilation=1
NVIDIA GeForce RTX 3090
20.2ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=256 · dilation=2
NVIDIA GeForce RTX 3090
21.0ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=256 · dilation=2
NVIDIA GeForce RTX 3090
21.1ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=256 · dilation=1
NVIDIA GeForce RTX 3090
21.9ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=4096 · dilation=1
NVIDIA GeForce RTX 3090
28.9ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=4096 · dilation=2
NVIDIA GeForce RTX 3090
29.1ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=512 · dilation=2
NVIDIA GeForce RTX 3090
39.9ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=512 · dilation=1
NVIDIA GeForce RTX 3090
42.2ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=512 · dilation=1
NVIDIA GeForce RTX 3090
42.7ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=512 · dilation=2
NVIDIA GeForce RTX 3090
44.5ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=1024 · dilation=1
NVIDIA GeForce RTX 3090
77.7ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=1024 · dilation=2
NVIDIA GeForce RTX 3090
78.0ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=1024 · dilation=2
NVIDIA GeForce RTX 3090
87.5ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=1024 · dilation=1
NVIDIA GeForce RTX 3090
88.0ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=2048 · dilation=2
NVIDIA GeForce RTX 3090
162.8ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=2048 · dilation=2
NVIDIA GeForce RTX 3090
169.9ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=2048 · dilation=1
NVIDIA GeForce RTX 3090
172.9ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=2048 · dilation=1
NVIDIA GeForce RTX 3090
181.8ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=4096 · dilation=2
NVIDIA GeForce RTX 3090
318.9ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=4096 · dilation=1
NVIDIA GeForce RTX 3090
346.6ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=4096 · dilation=2
NVIDIA GeForce RTX 3090
360.8ms
#2 of 2
2025-05-27
Fused neighborhood attentionfp32 · [512] · seq=4096 · dilation=1
NVIDIA GeForce RTX 3090
368.1ms
#2 of 2
2025-05-27
Reported · How evidence levels are derived →
Source and license
sourceavailable
revision digestsha256:0c3e4e2a4849358a1d5183daed7f408a3422f6cc411cc6bd7d4ebeead8e57e00
license declaredBSD-3-Clause
license concludedBSD-3-Clause
imported2026-08-20
Best evidence level for this revision: reported
JSON