ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-03 21:58:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

hot fix eid

Browse Source
This commit is contained in:
coderfeli
2025-02-20 10:18:01 +00:00
parent b6fbe4f7fe
commit cf9be76c73
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
include/ck/tensor_operation/gpu/grid/gridwise_moe_gemm.hpp
View File

@@ -1158,7 +1158,8 @@ struct GridwiseMoeGemm
// constexpr int expert_tile_cnt[8] = {2, 1, 1, 2, 2, 2, 1, 2};
// const index_t b_block_id = blockIdx.x % problem.NBlock;
const index_t expert_block_id = blockIdx.x / problem.NBlock;
const index_t expert_id = __builtin_amdgcn_readfirstlane(p_sorted_expert_ids[blockIdx.x / problem.NBlock]);
const index_t expert_id = NSwizzle ? __builtin_amdgcn_readfirstlane(p_sorted_expert_ids[blockIdx.x / problem.NBlock]) :
__builtin_amdgcn_readfirstlane(p_sorted_expert_ids[blockIdx.y]);
const auto block_mn = [&]() -> std::pair<int, int> {
if constexpr (NSwizzle)
{