ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-19 12:30:16 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit 'f211156ce6e9a8411c9ab8c3647147b6a9cf78d8' into develop

Browse Source
This commit is contained in:
assistant-librarian[bot]
2025-12-02 07:14:17 +00:00
parent 94dda8df22
commit ae8f3a3b19
5 changed files with 206 additions and 294 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
example/ck_tile/18_flatmm/mxgemm/mx_flatmm.cpp
View File

@@ -158,7 +158,7 @@ auto create_args(int argc, char* argv[])
.insert("stride_c", "0", "Tensor C stride")
.insert("v", "1", "0. No validation, 1. Validation on CPU, 2. Validation on GPU")
.insert(
"mx_prec", "fp4xfp4", "data type for activation and weight, support: fp6xfp6, fp8xfp8")
"mx_prec", "fp4xfp4", "data type for activation and weight, support: fp4xfp4, fp8xfp8")
.insert("warmup", "50", "number of iterations before benchmark the kernel")
.insert("repeat", "100", "number of iterations to benchmark the kernel")
.insert("timer", "gpu", "gpu:gpu timer, cpu:cpu timer")

2
example/ck_tile/18_flatmm/mxgemm/mx_flatmm_instance.hpp
View File

@@ -75,7 +75,7 @@ float mx_flatmm_calc(const ck_tile::ScaleFlatmmHostArgs<ScaleM, ScaleN>& args,
HasHotLoop,
TailNum>;
using MXFlatmmPipeline = ck_tile::MXF4FlatmmPipelineAGmemBGmemCRegV1<MXPipelineProblem>;
using MXFlatmmPipeline = ck_tile::MXFlatmmPipelineAGmemBGmemCRegV1<MXPipelineProblem>;
using TilePartitioner =
ck_tile::GemmSpatiallyLocalTilePartitioner<FlatmmShape,