Compare runs
cohort sha256:37651991c9fc2678… · source-native profile
ranking-v1
no ranks: not comparable
Comparison identity
operation
liger-fused-neighborhood-attention
workload
seq = 64 · batch = 2 · heads = 8 · hidden = 512 · dilation = 1 · kernel_size = 7 · fp32 · 37fb4d8f
protocol
Liger-Kernel benchmark scripts · 080e6e66
environment
NVIDIA H100 · 2750ae38
correctness policy
52ff5e65
metric
latency median (ns)
Context · 10 fields›
architecture
sm_90
CUDA
unknown
driver
unknown
framework
unknown
samples
unknown
evidence
reported
license
BSD-3-Clause
source
available
status
passed
observed
2025-05-27
No winner: these runs didn't measure the same thing. The rows above show what differs. Why comparable?
Add run ›
Export ›
Markdown CSV JSON
PyTorch · sha256:638908e34fcb7… · observed 2025-05-27