From c1aefb42bf0b1e8135f0c3976fd021465a04f093 Mon Sep 17 00:00:00 2001 From: qin letao Date: Wed, 2 Apr 2025 07:36:46 +0000 Subject: [PATCH] fix n pos in scale --- .../gpu/grid/gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp b/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp index 27818b6964..50eb43e51f 100644 --- a/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp +++ b/include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v3_b_scale.hpp @@ -1464,7 +1464,7 @@ struct GridwiseGemm_xdl_cshuffle_v3 1, false>( b_scale_grid_desc_bn_ak, - make_multi_index(block_n_id * NPerBlock / ScaleBlockN + b_thread_offset_n, + make_multi_index((block_n_id * NPerBlock + b_thread_offset_n) / ScaleBlockN, b_thread_offset_k / ScaleBlockK)); constexpr auto b_scale_thread_slice_copy_step =