ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-06-29 19:28:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

update kernel

Browse Source
This commit is contained in:
joye
2025-06-11 12:38:22 +08:00
parent 7bc604f06a
commit 4c85e10281
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
View File

@@ -586,8 +586,8 @@ __global__ void kernel_grouped_conv_bwd_data_optimized_v2(const ABDataType* __re
(rel_in_y + y) * TileInW * GroupPerBlockInFP4 +
(rel_in_x + x) * GroupPerBlockInFP4 + group_out_id];
ABDTypeVec_t shmem_k_vec = reinterpret_cast<ABDTypeVec_t*>(
shmem_k)[(kernel_y + y * up_h) * kernelW * GroupPerBlock +
(kernel_x + x * up_w) * GroupPerBlock + group_out_id];
shmem_k)[(kernel_y + y * up_h) * kernelW * GroupPerBlockInFP4 +
(kernel_x + x * up_w) * GroupPerBlockInFP4 + group_out_id];
static_for<0, ElementPerInFP4, 1>{}([&](auto idx) {
auto x_val = shmem_x_vec.template AsType<ABDataType>()[idx];