ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 10:09:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

Enable DPP8 GEMM on Navi3 (#892)

Browse Source 
[ROCm/composable_kernel commit: 8f84a01237]
This commit is contained in:
Bartlomiej Wroblewski
2023-09-08 18:14:57 +02:00
committed by GitHub
parent c2866bb432
commit bf5b711799
2 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
include/ck/tensor_operation/gpu/device/impl/device_gemm_dpp.hpp
View File

@@ -168,7 +168,8 @@ struct DeviceGemmDpp : public DeviceGemm<ALayout,
static bool IsSupportedArgument(const Argument& karg)
{
if(ck::get_device_name() == "gfx1030")
if(ck::get_device_name() == "gfx1030" || ck::get_device_name() == "gfx1100" ||
ck::get_device_name() == "gfx1101" || ck::get_device_name() == "gfx1102")
{
return GridwiseGemm::CheckValidity(karg);
}

3
include/ck/tensor_operation/gpu/grid/gridwise_gemm_dpp.hpp
View File

@@ -28,7 +28,8 @@ __global__ void
#endif
kernel_gemm_dpp(const typename GridwiseGemm::Argument karg)
{
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1030__))
#if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx1030__) || defined(__gfx1100__) || \
defined(__gfx1101__) || defined(__gfx1102__))
__shared__ char p_shared[GridwiseGemm::GetSharedMemoryNumberOfByte()];
const auto a_grid_desc_ak0_m_ak1 = amd_wave_read_first_lane(