Support bf16 in blackwell cutlass decode attention kernel #5712
fbgemm_gpu_benchmark_rocm.yml
on: pull_request
Matrix: build_artifact
Matrix: benchmark_artifact
Artifacts
Produced during runtime
Name | Size | Digest | |
---|---|---|---|
fbgemm_gpu_nightly_rocm_x86_gcc_py3.13_rocm6.3.whl
|
246 MB |
sha256:6a1e93420ffb2d6f1ed27fa5483ba46d6c5e0daa603b0ebf1d2eac576b6125d3
|
|
fbgemm_gpu_traces_x86_gcc_py3.13_rocm6.3.zip
|
5.94 MB |
sha256:5485267c5fbdcf17d971fa34c9dc532da2a00ed74099fee21ae2fc129c02062a
|
|