ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 05:01:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix grouped conv bwd data tests on gfx950 (#2173)

Browse Source
This commit is contained in:
jefyang1
2025-05-09 09:01:06 -07:00
committed by GitHub
parent a23390163d
commit 6b1a339b6f
1 changed files with 2 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp
View File

@@ -179,8 +179,7 @@ __global__ void
const ComputePtrOffsetOfN compute_ptr_offset_of_n,
const index_t num_k_per_block)
{
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__) || \
defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__))
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx9__))
// offset base pointer for each work-group
const index_t g_idx = __builtin_amdgcn_readfirstlane(blockIdx.z);
const index_t n_idx = __builtin_amdgcn_readfirstlane(blockIdx.y / karg.KBatch);
@@ -251,8 +250,7 @@ __global__ void
const ComputePtrOffsetOfN compute_ptr_offset_of_n,
const index_t num_k_per_block)
{
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__) || \
defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__))
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx9__))
const index_t g_idx = __builtin_amdgcn_readfirstlane(blockIdx.z);
const index_t n_idx = __builtin_amdgcn_readfirstlane(blockIdx.y / karg.KBatch);
const index_t k_idx =