[llvm] [X86] shouldReduceLoadWidth - don't split loads if ANY uses are a extract+store or a full width legal binop (PR #129695)

[Phoebe Wang via llvm-commits]

================
@@ -4228,7 +4228,7 @@ define <4 x float> @uitofp_load_4i64_to_4f32(ptr%a) {
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vmovdqa (%rdi), %ymm0
 ; AVX1-NEXT:    vpsrlq $1, %xmm0, %xmm1
-; AVX1-NEXT:    vmovdqa 16(%rdi), %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
----------------
phoebewang wrote:

Considering data is already in cache, is vextractf128 better than vmovdqa? In any targets?

https://github.com/llvm/llvm-project/pull/129695