ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-19 20:40:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge commit '4dcc3e59c1c0195dae7ee9da9ab76d18a4cafe9f' into develop

Browse Source
This commit is contained in:
assistant-librarian[bot]
2025-12-11 17:17:01 +00:00
parent b69f9eb589
commit f9ad462542
44 changed files with 1303 additions and 170 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
tutorial/ck_tile/01_naive_gemm/host_level/practice_gemm_host_pipeline_agmem_bgmem_creg.hpp
View File

@@ -28,9 +28,9 @@ struct PracticeGemmHostPipeline
{
// Size of the entire problem
const auto M = a_dram.get_tensor_descriptor().get_length(number<0>{}); // M x K
const auto N = c_dram_ref.get_tensor_descriptor().get_length(number<1>{}); // M x N
const auto K = a_dram.get_tensor_descriptor().get_length(number<1>{}); // M x K
const auto M = a_dram.get_tensor_descriptor().get_length(number<0>{}); // M x K
const auto N = c_dram.get_tensor_descriptor().get_length(number<1>{}); // M x N
const auto K = a_dram.get_tensor_descriptor().get_length(number<1>{}); // M x K
// Size of the block tile
const auto MPerBlock = BlockTile::at(number<0>{});
@@ -83,7 +83,7 @@ struct PracticeGemmHostPipeline
__shared__ char p_smem_char[block_gemm_pipeline.GetStaticLDSSize()];
const auto c_block_tile =
block_gemm_pipeline(a_block_window, b_block_window, num_loops_k, p_smem_char);
auto c_window = make_tile_window(c_dram_ref,
auto c_window = make_tile_window(c_dram,
make_tuple(number<MPerBlock>{}, number<NPerBlock>{}),
{tile_origin_m, tile_origin_n});
store_tile(c_window, c_block_tile);