[llvm] [X86] Support lowering of FMINIMUMNUM/FMAXIMUMNUM (PR #121464)

[Matt Arsenault via llvm-commits]

================
@@ -15,7 +15,11 @@ entry:
 define <2 x double> @max_nnan_v2f64(<2 x double> %a, <2 x double> %b) {
 ; AARCH64-LABEL: max_nnan_v2f64:
 ; AARCH64:       // %bb.0: // %entry
-; AARCH64-NEXT:    fmaxnm v0.2d, v0.2d, v1.2d
+; AARCH64-NEXT:    mov d2, v1.d[1]
+; AARCH64-NEXT:    mov d3, v0.d[1]
+; AARCH64-NEXT:    fmaxnm d0, d0, d1
----------------
arsenm wrote:

Looks similar to the AMDGPU regression, this is now full scalarization 

https://github.com/llvm/llvm-project/pull/121464