leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 26 Packages Projects Releases Wiki Activity

fMHA: Add backward pass (#844)

Browse Source 
* fMHA: Add backward pass

* Better checks for strides/alignments

* Remove fb-internal URL

* torch.Tensor.untyped_storage requires pytorch 2.0+

* minor changes

* make test

---------

Co-authored-by: danthe3rd <danthe3rd>
Co-authored-by: Haicheng Wu <haichengw@nvidia.com>
This commit is contained in:
dan_the_3rd
2023-04-06 20:44:58 -04:00
committed by GitHub
co-authored by danthe3rd <danthe3rd> Haicheng Wu
parent e2d439ee7e
commit 9b8166e3f0
9 changed files with 2931 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/library/CMakeLists.txt
+5
View File
@@ -30,6 +30,11 @@ include(GNUInstallDirs)
find_package(Python3 3.5 COMPONENTS Interpreter REQUIRED)
# Set Python3_EXECUTABLE to be visible from global scope.
# In CMake 3.24, this could be supported by adding the GLOBAL field
# to find_package above (https://cmake.org/cmake/help/latest/command/find_package.html#id7)
set(Python3_EXECUTABLE ${Python3_EXECUTABLE} CACHE INTERNAL "Path to python3 executable")
add_library(cutlass_library_includes INTERFACE)
add_library(nvidia::cutlass::library::includes ALIAS cutlass_library_includes)
set_target_properties(cutlass_library_includes PROPERTIES EXPORT_NAME library::includes)