Skip to content

[SYCL][CUDA][libclc] Add bitwise group collective functions for remaining integral types #6405

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jul 8, 2022
Merged

[SYCL][CUDA][libclc] Add bitwise group collective functions for remaining integral types #6405

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f4a9efe
added bitwise group collective fucntions for remainning integral types
t4c1 Jul 6, 2022
030d39b
do not use redux for long
t4c1 Jul 7, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 42 additions & 0 deletions libclc/ptx-nvidiacl/libspirv/group/collectives.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -249,13 +249,34 @@ __CLC_SUBGROUP_COLLECTIVE(FMax, __CLC_MAX, half, -HALF_MAX)
__CLC_SUBGROUP_COLLECTIVE(FMax, __CLC_MAX, float, -FLT_MAX)
__CLC_SUBGROUP_COLLECTIVE(FMax, __CLC_MAX, double, -DBL_MAX)

__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, uchar, ~0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, uchar, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, uchar, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, char, ~0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, char, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, char, 0)

__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, ushort, ~0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, ushort, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, ushort, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, short, ~0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, short, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, short, 0)

__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, uint, ~0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, uint, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, uint, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, int, ~0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, int, 0)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, int, 0)

__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, ulong, ~0l)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, ulong, 0l)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, ulong, 0l)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseAndKHR, __CLC_AND, and, long, ~0l)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseOrKHR, __CLC_OR, or, long, 0l)
__CLC_SUBGROUP_COLLECTIVE_REDUX(BitwiseXorKHR, __CLC_XOR, xor, long, 0l)

#undef __CLC_SUBGROUP_COLLECTIVE_BODY
#undef __CLC_SUBGROUP_COLLECTIVE
#undef __CLC_SUBGROUP_COLLECTIVE_REDUX
Expand Down Expand Up @@ -376,13 +397,34 @@ __CLC_GROUP_COLLECTIVE(FMax, __CLC_MAX, half, -HALF_MAX)
__CLC_GROUP_COLLECTIVE(FMax, __CLC_MAX, float, -FLT_MAX)
__CLC_GROUP_COLLECTIVE(FMax, __CLC_MAX, double, -DBL_MAX)

__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, uchar, ~0)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, uchar, 0)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, uchar, 0)
__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, char, ~0)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, char, 0)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, char, 0)

__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, ushort, ~0)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, ushort, 0)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, ushort, 0)
__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, short, ~0)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, short, 0)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, short, 0)

__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, uint, ~0)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, uint, 0)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, uint, 0)
__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, int, ~0)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, int, 0)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, int, 0)

__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, ulong, ~0l)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, ulong, 0l)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, ulong, 0l)
__CLC_GROUP_COLLECTIVE(BitwiseAndKHR, __CLC_AND, long, ~0l)
__CLC_GROUP_COLLECTIVE(BitwiseOrKHR, __CLC_OR, long, 0l)
__CLC_GROUP_COLLECTIVE(BitwiseXorKHR, __CLC_XOR, long, 0l)

// half requires additional mangled entry points
_CLC_DEF _CLC_CONVERGENT half _Z17__spirv_GroupFAddjjDF16_(uint scope, uint op,
half x) {
Expand Down