[clang] [X86][Clang] VectorExprEvaluator::VisitCallExpr / InterpretBuiltin - Allow SSE/AVX FP MAX/MIN intrinsics to be used in constexpr (PR #171966)

Simon Pilgrim via cfe-commits cfe-commits at lists.llvm.org
Fri Jan 9 14:21:06 PST 2026


================
@@ -387,9 +385,8 @@ _mm_min_ps(__m128 __a, __m128 __b)
 /// \returns A 128-bit vector of [4 x float] whose lower 32 bits contain the
 ///    maximum value between both operands. The upper 96 bits are copied from
 ///    the upper 96 bits of the first source operand.
-static __inline__ __m128 __DEFAULT_FN_ATTRS
-_mm_max_ss(__m128 __a, __m128 __b)
-{
+static __inline__ __m128 __DEFAULT_FN_ATTRS_CONSTEXPR _mm_max_ss(__m128 __a,
+                                                                 __m128 __b) {
----------------
RKSimon wrote:

scalar not supported - undo

https://github.com/llvm/llvm-project/pull/171966


More information about the cfe-commits mailing list