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

conv:tf32:add missed instances (#3081)

Browse Source 
* conv:tf32:add missed instances
This commit is contained in:
yinglu
2025-10-24 16:28:36 +08:00
committed by GitHub
parent d0364641ed
commit 6bbc05e1bd
35 changed files with 1331 additions and 120 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_abd_xdl_cshuffle.hpp
View File

@@ -1789,28 +1789,6 @@ struct DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
}
}
if constexpr(is_same_v<AComputeDataType, ck::tf32_t> ||
is_same_v<BComputeDataType, ck::tf32_t>)
{
if(!(ck::get_device_name() == "gfx942"))
{
if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
{
std::cout << "TF32 is enabled on gfx942 only" << std::endl;
}
return false;
}
if constexpr(!is_same_v<AComputeDataType, BComputeDataType>)
{
if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
{
std::cout << "ComputeDataType for A and B should be same while using TF32"
<< std::endl;
}
return false;
}
}
return false;
}