[llvm] [LV][AArch64] LoopVectorizer allows scalable frem instructions (PR #76247)

[Maciej Gabka via llvm-commits]

================
@@ -137,14 +138,14 @@ define void @fdiv() {
 
 define void @frem() {
 ; CHECK-LABEL: 'frem'
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V4F16 = frem <vscale x 4 x half> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V8F16 = frem <vscale x 8 x half> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V16F16 = frem <vscale x 16 x half> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V2F32 = frem <vscale x 2 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V4F32 = frem <vscale x 4 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V8F32 = frem <vscale x 8 x float> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V2F64 = frem <vscale x 2 x double> undef, undef
-; CHECK-NEXT:  Cost Model: Invalid cost for instruction: %V4F64 = frem <vscale x 4 x double> undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %V4F16 = frem <vscale x 4 x half> undef, undef
----------------
mgabka wrote:

in these tests veclib is not available so the cost should still be invalid

https://github.com/llvm/llvm-project/pull/76247