leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 30 Packages Projects Releases Wiki Activity

CUTLASS 3.2.1 (#1113)

Browse Source 
* Updates for 3.2.1 release.

* Minor fix in gemm op profiler for raster order.

* Add scheduler mapping for raster order in the kernels.
This commit is contained in:
ANIKET SHIVAM
2023-09-26 17:24:26 -04:00
committed by GitHub
parent e0aaa3c3b3
commit 90d3b0fb18
428 changed files with 22252 additions and 21761 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/cute/swizzle_ptr.hpp
+6 -3
View File
@@ -70,6 +70,8 @@ struct smem_ptr_swizzle
{
static_assert(is_empty<Swizzle>::value, "Swizzle can't have state.");
static const uint32_t ElementsPerStoredItem = sizeof(T) * 8 / sizeof_bits_v<T>;
CUTE_HOST_DEVICE constexpr
T* get() const
{
@@ -98,6 +100,7 @@ struct smem_ptr_swizzle
CUTE_HOST_DEVICE constexpr
T& operator[](Int const& i) const
{
static_assert(sizeof_bits_v<T> >= 8, "Use subbyte_iterator to access the element");
return *apply_swizzle(get() + i);
}
@@ -105,7 +108,7 @@ struct smem_ptr_swizzle
CUTE_HOST_DEVICE constexpr
smem_ptr_swizzle operator+(Int const& i) const
{
return {ptr_ + i};
return {ptr_ + i / ElementsPerStoredItem};
}
T* ptr_;
@@ -286,14 +289,14 @@ CUTE_HOST_DEVICE void print(smem_ptr_flag_bits<B> const& ptr)
template <class T, int B, int M, int S>
CUTE_HOST_DEVICE void print(smem_ptr_swizzle<T,Swizzle<B,M,S>> const& ptr)
{
printf("smem_ptr_S<%d,%d,%d>_%db(%p)", B, M, S, int(8*sizeof(T)), ptr.get());
printf("smem_ptr_S<%d,%d,%d>_%db(%p)", B, M, S, int(sizeof_bits<T>::value), ptr.get());
}
#if !defined(__CUDACC_RTC__)
template <class T, int B, int M, int S>
CUTE_HOST std::ostream& operator<<(std::ostream& os, smem_ptr_swizzle<T,Swizzle<B,M,S>> const&)
{
return os << "smem_ptr_S<" << B << "," << M << "," << S << ">_" << int(8*sizeof(T)) << "b";
return os << "smem_ptr_S<" << B << "," << M << "," << S << ">_" << int(sizeof_bits<T>::value) << "b";
}
#endif