| Issue |
152324
|
| Summary |
[Headers][X86] Allow AVX512 integer reduction intrinsics to be used in constexpr
|
| Labels |
backend:X86,
clang:headers,
constexpr
|
| Assignees |
|
| Reporter |
RKSimon
|
The AVX512 `_mm512_reduce_*_epi*` intrinsic wrap the builtin reduction intrinsics which can already be used in constexpr:
```
_mm512_reduce_add_epi32 _mm512_reduce_add_epi64
_mm512_reduce_mul_epi32 _mm512_reduce_mul_epi64
_mm512_reduce_and_epi32 _mm512_reduce_and_epi64
_mm512_reduce_or_epi32 _mm512_reduce_or_epi64
_mm512_reduce_max_epi32 _mm512_reduce_max_epi64
_mm512_reduce_min_epi32 _mm512_reduce_min_epi64
_mm512_reduce_max_epu32 _mm512_reduce_max_epu64
_mm512_reduce_min_epu32 _mm512_reduce_min_epu64
```
(ignore the AVX512 mask/maskz variants for now)
You can use https://github.com/llvm/llvm-project/pull/152305 as reference.
![]()
_______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
