Compare runs
cohort sha256:de3688cbc19f579e… · source-native profile
ranking-v1
no ranks: not comparable
Comparison identity
operation
gqa-paged-decode-h32-kv4-d128-ps1
workload
num_pages = 52831 · batch_size = 64 · len_indptr = 65 · num_kv_indices = 52815 · bf16 · ed0bcb93
protocol
flashinfer-bench · 7d44d206
environment
NVIDIA B200 · f4292a8b
correctness policy
64fb869e
metric
latency mean (ns)
Context · 10 fields›
architecture
sm_100
CUDA
12.8
driver
unknown
framework
pytorch 2.8.0+cu128
samples
unknown
evidence
reported
license
Apache-2.0
source
available
status
passed
observed
2025-10-21
No winner: these runs didn't measure the same thing. The rows above show what differs. Why comparable?
Add run ›
Export ›
Markdown CSV JSON
flashinfer_wrapper_78fd04 · sha256:0c72ec9c8fdbd… · observed 2025-10-21