CUTLASS 2.10 updates (#622)

Co-authored-by: Aniket Shivam <ashivam@nvidia.com>
2022-09-12 18:26:30 -07:00
parent beae168f90
commit e773429f7e
96 changed files with 8365 additions and 1667 deletions
--- a/include/cutlass/arch/memory.h
+++ b/include/cutlass/arch/memory.h
@@ -69,7 +69,7 @@ struct global_load;
 /////////////////////////////////////////////////////////////////////////////////////////////////

 // The redundant mov PTX instruction is used to enforce the compiler to
-// initialize data to zero before ld.global
+// keep the initializing code before ld.global
 template <typename AccessType>
 struct global_load<AccessType,
                   32