ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-02 12:41:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

Handle moe_fp8 no-mainloop cases. Supprese no-mainloop check (#2438)

Browse Source 
Co-authored-by: felix <felix.li@amd.com>
This commit is contained in:
huaiguxu
2025-07-16 15:44:34 +08:00
committed by GitHub
parent 3499fe67ff
commit c1badfd30c
2 changed files with 45 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
include/ck/tensor_operation/gpu/device/impl/device_moe_gemm.hpp
View File

@@ -325,12 +325,50 @@ struct DeviceMoeGemm : public DeviceGemmMultipleDSplitKBPreShuffle<ALayout,
// Tail number always 1
if constexpr(BlkGemmPipelineVer == BlockGemmPipelineVersion::v1)
{
const auto kernel = kernel_moe_gemm<GridwiseGemm,
true,
InMemoryDataOperationEnum::Set,
minimum_occupancy,
TailNumber::Odd>;
RunKernel(kernel);
if(GridwiseGemm::CalculateKBlockLoopTailNum(K_split) == TailNumber::Odd)
{
const auto kernel = kernel_moe_gemm<GridwiseGemm,
false,
MemoryDataOp,
minimum_occupancy,
TailNumber::Odd>;
RunKernel(kernel);
}
else
{
const auto kernel = kernel_moe_gemm<GridwiseGemm,
false,
MemoryDataOp,
minimum_occupancy,
TailNumber::Even>;
RunKernel(kernel);
}
}
else if constexpr(BlkGemmPipelineVer == BlockGemmPipelineVersion::v2 ||
BlkGemmPipelineVer == BlockGemmPipelineVersion::v3)
{
if(GridwiseGemm::CalculateKBlockLoopTailNum(K_split) == TailNumber::Odd)
{
const auto kernel = kernel_moe_gemm_2lds<GridwiseGemm,
false,
MemoryDataOp,
minimum_occupancy,
TailNumber::Odd>;
RunKernel(kernel);
}
else
{
const auto kernel = kernel_moe_gemm_2lds<GridwiseGemm,
false,
MemoryDataOp,
minimum_occupancy,
TailNumber::Even>;
RunKernel(kernel);
}
}
else
{
throw std::runtime_error("todo: only v1 & v2 support now");
}
}
#endif