ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-21 21:39:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

adding ConstantMergedTensorDescriptor, refactering ConstantTensorDescriptor, Sequence

Browse Source
This commit is contained in:
Chao Liu
2019-05-21 16:17:58 -05:00
parent cd29b09a82
commit acd7082fe1
38 changed files with 1238 additions and 768 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/include/conv_common.hip.hpp
View File

@@ -30,7 +30,7 @@ __host__ __device__ constexpr auto get_convolution_output_default_4d_tensor_desc
constexpr auto HO = HI + 1 - Y;
constexpr auto WO = WI + 1 - X;
return make_ConstantTensorDescriptor(Sequence<N, K, HO, WO>{});
return make_packed_ConstantTensorDescriptor(Sequence<N, K, HO, WO>{});
}
template <class InDesc, class WeiDesc, class LowerPads, class UpperPads>
@@ -67,7 +67,7 @@ __host__ __device__ constexpr auto get_convolution_with_padding_output_default_4
constexpr auto HO = HI + HPadLow + HPadUp + 1 - Y;
constexpr auto WO = WI + WPadLow + WPadUp + 1 - X;
return make_ConstantTensorDescriptor(Sequence<N, K, HO, WO>{});
return make_packed_ConstantTensorDescriptor(Sequence<N, K, HO, WO>{});
}
template <class InDesc, class WeiDesc, class OutDesc>