Improve implementation of __masked_store_blend_64() for AVX target by doing two 8-wide 32-bit blends rather than serializing. Fixes issue #29
This commit is contained in:
1
.gitignore
vendored
1
.gitignore
vendored
@@ -4,3 +4,4 @@ depend
|
||||
ispc
|
||||
ispc_test
|
||||
objs
|
||||
docs/doxygen
|
||||
|
||||
Reference in New Issue
Block a user