ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-12 17:26:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

tidy up

Browse Source
This commit is contained in:
Chao Liu
2019-04-02 20:30:00 -05:00
parent e6c86f81b5
commit e2313c9eca
3 changed files with 11 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
src/include/common.hip.hpp
View File

@@ -62,10 +62,9 @@ __host__ __device__ constexpr T min(T x, Ts... xs)
return x < y ? x : y;
}
}// namespace mod_conv
} // namespace mod_conv
#if DEVICE_BACKEND_HIP
// cast a pointer of LDS to its address
extern "C" __attribute__((address_space(3))) void* __to_local(void* p)[[hc]];
#endif

8
src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn.hip.hpp
View File

@@ -287,10 +287,10 @@ struct GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn
#elif 0
blockwise_gemm.Run_asm
#endif
(p_wei_block + wei_cyxk_block_desc.Get1dIndex(0, y, x, 0),
p_in_block + y * Wi + x,
p_out_thread,
f_accum);
(p_wei_block + wei_cyxk_block_desc.Get1dIndex(0, y, x, 0),
p_in_block + y * Wi + x,
p_out_thread,
f_accum);
}
}
}

9
src/include/gridwise_convolution_implicit_gemm_v2_chwn_cyxk_khwn_lds_double_buffer.hip.hpp
View File

@@ -36,11 +36,14 @@ template <index_t GridSize,
index_t WeiBlockCopyDataPerRead>
struct GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn_lds_double_buffer
{
__host__ __device__ constexpr GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn_lds_double_buffer() {}
__host__
__device__ constexpr GridwiseConvolutionImplicitGemm_v2_chwn_cyxk_khwn_lds_double_buffer()
{
}
__device__ void Run(const Float* const __restrict__ p_in_global,
const Float* const __restrict__ p_wei_global,
Float* const __restrict__ p_out_global) const
const Float* const __restrict__ p_wei_global,
Float* const __restrict__ p_out_global) const
{
constexpr auto I0 = Number<0>{};
constexpr auto I1 = Number<1>{};