[llvm] [AArch64][SVE] Use SVE for scalar FP converts in streaming[-compatible] functions (PR #112564)

[Benjamin Maxwell via llvm-commits]

================
@@ -3072,9 +3073,10 @@ define float @ucvtf_i32_f32(ptr %0) {
 define double @ucvtf_i32_f64(ptr %0) {
 ; CHECK-LABEL: ucvtf_i32_f64:
 ; CHECK:       // %bb.0:
+; CHECK-NEXT:    ldr w8, [x0]
 ; CHECK-NEXT:    ptrue p0.d
-; CHECK-NEXT:    ldr s0, [x0]
-; CHECK-NEXT:    ucvtf z0.d, p0/m, z0.s
+; CHECK-NEXT:    fmov d0, x8
+; CHECK-NEXT:    ucvtf z0.d, p0/m, z0.d
 ; CHECK-NEXT:    // kill: def $d0 killed $d0 killed $z0
----------------
MacDue wrote:

Now loads to GPR instead of FPR (possible hazard + extra fmov)

https://github.com/llvm/llvm-project/pull/112564