ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-18 01:28:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

generalize the pipeline scheduling.

Browse Source
This commit is contained in:
aska-0096
2025-05-16 10:41:59 +00:00
committed by Ding, Yi
parent 3e8b07ef58
commit e2c8f98fef
3 changed files with 90 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
example/67_gemm_microscaling/gemm_mx_fp4.cpp
View File

@@ -50,14 +50,14 @@ using DeviceOpInstance = ck::tensor_operation::device::DeviceGemmMX_Xdl_CShuffle
GemmSpec, // GemmSpec
ScaleBlockSize, // ScaleBlockSize: Scaling block size
256, // BlockSize: Thread block size
128, // MPerBlock
192, // MPerBlock
256, // NPerBlock
KPerBlock, // KPerBlock
16, // AK1
16, // BK1
16, // MPerXDL
16, // NPerXDL
4, // MXdlPerWave
6, // MXdlPerWave
8, // NXdlPerWave
S<8, 32, 1>, // ABlockTransferThreadClusterLengths_AK0_M_AK1
S<1, 0, 2>, // ABlockTransferThreadClusterArrangeOrder