leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 27 Packages Projects Releases Wiki Activity

Fix typos 2 (#842)

Browse Source 
Co-authored-by: Haicheng Wu <57973641+hwu36@users.noreply.github.com>
This commit is contained in:
Alexander Pivovarov
2023-03-09 23:22:56 -05:00
committed by GitHub
co-authored by Haicheng Wu
parent c4f6b8c6bc
commit 7e370c9637
161 changed files with 310 additions and 309 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/cutlass/gemm/device/ell_gemm.h
+1 -1
View File
@@ -547,7 +547,7 @@ public:
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm.h
+1 -1
View File
@@ -521,7 +521,7 @@ public:
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_array.h
+1 -1
View File
@@ -476,7 +476,7 @@ public:
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_batched.h
+1 -1
View File
@@ -454,7 +454,7 @@ public:
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_complex.h
+1 -1
View File
@@ -475,7 +475,7 @@ public:
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_layernorm_mainloop_fusion.h
+1 -1
View File
@@ -194,7 +194,7 @@ class GemmLayernormMainloopFusion :
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_universal.h
+1 -1
View File
@@ -219,7 +219,7 @@ class GemmUniversal :
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_universal_with_broadcast.h
+1 -1
View File
@@ -198,7 +198,7 @@ class GemmUniversalWithBroadcast :
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/gemm_with_k_reduction.h
+1 -1
View File
@@ -211,7 +211,7 @@ class GemmWithKReduction :
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/rank_2k.h
+1 -1
View File
@@ -348,7 +348,7 @@ public:
};
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchange operand.
/// Partial specialization for column-major output exchange operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/rank_k.h
+1 -1
View File
@@ -325,7 +325,7 @@ public:
};
////////////////////////////////////////////////////////////////////////////////
/// Parital specialization for column-major output exchange operand.
/// Partial specialization for column-major output exchange operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/symm.h
+1 -1
View File
@@ -408,7 +408,7 @@ public:
call GEMM mainloop for with RowMajor efficient-epilogue
********************************************************************************************************/
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/device/trmm.h
+1 -1
View File
@@ -563,7 +563,7 @@ For the mainloop and trmm kernel, `A` and `B` points to left-side and right-side
call GEMM mainloop for with RowMajor efficient-epilogue
********************************************************************************************************/
/// Parital specialization for column-major output exchanges problem size and operand.
/// Partial specialization for column-major output exchanges problem size and operand.
template <
/// Element type for A matrix operand
typename ElementA_,
include/cutlass/gemm/kernel/default_gemm_with_broadcast.h
+1 -1
View File
@@ -137,7 +137,7 @@ struct DefaultGemmWithBroadcast {
/////////////////////////////////////////////////////////////////////////////////////////////////
/// Parital specialization: ArchTag = cutlass::arch::Sm70
/// Partial specialization: ArchTag = cutlass::arch::Sm70
///
///
template <
include/cutlass/gemm/kernel/default_gemm_with_reduction.h
+1 -1
View File
@@ -138,7 +138,7 @@ struct DefaultGemmWithReduction {
/////////////////////////////////////////////////////////////////////////////////////////////////
/// Parital specialization: ArchTag = cutlass::arch::Sm70
/// Partial specialization: ArchTag = cutlass::arch::Sm70
///
///
template <
include/cutlass/gemm/kernel/rank_2k_grouped_problem_visitor.h
+1 -1
View File
@@ -138,7 +138,7 @@
i = i_macro
j = j_macro
Handling cases with grid dimensions that aren't multiples of eachother
Handling cases with grid dimensions that aren't multiples of each other
----------------------------------------------------------------------
Even though threadblock shapes M and N are typically multiples of one another, the grid
for a given problem may not have dimensions of the same ratio as that of the threadblock.
include/cutlass/gemm/kernel/sm90_gemm_tma.hpp
+1 -1
View File
@@ -196,7 +196,7 @@ public:
// Any Tensor Op MMA Atom in the WGMMA ISA is arch conditional to sm90a.
#if ! defined(__CUDA_ARCH_FEAT_SM90_ALL)
if constexpr(size<0>(typename TiledMma::AtomShape_MNK{}) == 64) {
printf("ERROR : Arch conditional MMA instruction used without targetting sm90a compute capability. Aborting.\n");
printf("ERROR : Arch conditional MMA instruction used without targeting sm90a compute capability. Aborting.\n");
return;
}
#endif
include/cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized.hpp
+1 -1
View File
@@ -186,7 +186,7 @@ public:
// Any Tensor Op MMA Atom in the WGMMA ISA is arch conditional to sm90a.
#if ! defined(__CUDA_ARCH_FEAT_SM90_ALL)
if constexpr(size<0>(typename TiledMma::AtomShape_MNK{}) == 64) {
printf("ERROR : Arch conditional MMA instruction used without targetting sm90a compute capability. Aborting.\n");
printf("ERROR : Arch conditional MMA instruction used without targeting sm90a compute capability. Aborting.\n");
return;
}
#endif
include/cutlass/gemm/kernel/sm90_gemm_tma_warpspecialized_persistent.hpp
+1 -1
View File
@@ -258,7 +258,7 @@ public:
// Any Tensor Op MMA Atom in the WGMMA ISA is arch conditional to sm90a.
#if ! defined(__CUDA_ARCH_FEAT_SM90_ALL)
if constexpr(size<0>(typename TiledMma::AtomShape_MNK{}) == 64) {
printf("ERROR : Arch conditional MMA instruction used without targetting sm90a compute capability. Aborting.\n");
printf("ERROR : Arch conditional MMA instruction used without targeting sm90a compute capability. Aborting.\n");
return;
}
#endif
include/cutlass/gemm/threadblock/ell_mma_pipelined.h
+1 -1
View File
@@ -271,7 +271,7 @@ public:
}
// Issue loads during the first warp-level matrix multiply-add *AFTER* issuing
// shared memory loads (which have the tighest latency requirement).
// shared memory loads (which have the tightest latency requirement).
//
// Mainloop
include/cutlass/gemm/threadblock/mma_planar_complex_pipelined.h
+1 -1
View File
@@ -321,7 +321,7 @@ public:
iterator_B_imag.clear_mask(gemm_k_iterations <= 1);
// Issue loads during the first warp-level matrix multiply-add *AFTER* issuing
// shared memory loads (which have the tighest latency requirement).
// shared memory loads (which have the tightest latency requirement).
//
// Mainloop