adding implicit gemm v3

This commit is contained in:
Chao Liu
2019-05-23 22:10:40 -05:00
parent 8a4b59785b
commit 1cc683a3a3
16 changed files with 347 additions and 95 deletions

View File

@@ -26,6 +26,11 @@ struct ConstantMergedTensorDescriptor
// TODO: check there is no duplication in OriginalDimMergeSeqs
}
__host__ __device__ static constexpr auto GetOriginalTensorDescriptor()
{
return OriginalTensorDesc{};
}
__host__ __device__ static constexpr index_t GetNumOfDimension() { return nDim; }
__host__ __device__ static constexpr index_t GetNumOfOriginalDimension()
@@ -120,3 +125,9 @@ __host__ __device__ constexpr auto make_ConstantMergedTensorDescriptor(OriginalT
{
return ConstantMergedTensorDescriptor<OriginalTensorDesc, OriginalDimMergeSeqs...>{};
}
template <class TDesc>
__host__ __device__ void print_ConstantMergedTensorDescriptor(TDesc, const char* s)
{
print_ConstantTensorDescriptor(TDesc::GetOriginalTensorDescriptor(), s);
}