ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-18 01:28:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add validity checks for MoE FlatMM scatter and enable bf16 hardware atomic

Browse Source
This commit is contained in:
Mohsen Saffari
2025-11-19 14:02:24 +00:00
parent ad57f6ef0b
commit 5bfda1f662
3 changed files with 11 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
example/ck_tile/18_flatmm/run_moe_flatmm_example.inc
View File

@@ -304,7 +304,7 @@ int run_moe_gemm_example_with_layouts(int argc,
const float max_accumulated_value =
*std::max_element(c_m_n_host_ref.mData.begin(), c_m_n_host_ref.mData.end());
const auto rtol_atol = calculate_rtol_atol<ADataType, BDataType, AccDataType, CDataType>(
[[maybe_unused]] const auto rtol_atol = calculate_rtol_atol<ADataType, BDataType, AccDataType, CDataType>(
K, 1 /*kbatch*/, max_accumulated_value);
c_m_n_ref_buf->FromDevice(c_m_n_host_ref.data());