ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-20 06:49:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

Disable SMFMA for gfx90a (#2182)

Browse Source
This commit is contained in:
Khushbu Agarwal
2025-05-09 00:18:07 -07:00
committed by GitHub
parent 3448e12609
commit ef72a4b9bc
2 changed files with 14 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tile_engine/ops/gemm/gemm_instance_builder.py
View File

@@ -535,7 +535,11 @@ struct GemmDispatcher {
((tile[6] == 32 and tile[7] == 32 and tile[8] == 16) or
(tile[6] == 16 and tile[7] == 16 and tile[8] == 32))
content += f"""
run_kernel<{group}::GemmKernel<{tile[0]}, {tile[1]}, {tile[2]}, {tile[3]}, {tile[4]}, {tile[5]}, {tile[6]}, {tile[7]}, {tile[8]}, {BOOL_MAP(sparse)}>>(c_m_n_dev_buf, c_m_n_host_result, c_m_n_dev_result, verify, args, stream);"""
#if defined(__gfx908__)
run_kernel<{group}::GemmKernel<{tile[0]}, {tile[1]}, {tile[2]}, {tile[3]}, {tile[4]}, {tile[5]}, {tile[6]}, {tile[7]}, {tile[8]}, {BOOL_MAP(False)}>>(c_m_n_dev_buf, c_m_n_host_result, c_m_n_dev_result, verify, args, stream);
#else
run_kernel<{group}::GemmKernel<{tile[0]}, {tile[1]}, {tile[2]}, {tile[3]}, {tile[4]}, {tile[5]}, {tile[6]}, {tile[7]}, {tile[8]}, {BOOL_MAP(sparse)}>>(c_m_n_dev_buf, c_m_n_host_result, c_m_n_dev_result, verify, args, stream);
#endif"""
content += f"""
}} else {{"""
for tile in tile_params: