[llvm] [AMDGPU] Add getCmpSelInstrCost override to re-enable SimplifyCFG speculation for vector types (PR #208043)

[Frederik Harwath via llvm-commits]

================
@@ -8,7 +8,6 @@ define i32 @reduce_i1(i32 %arg) {
 ; ALL-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i1 @llvm.vector.reduce.and.v1i1(<1 x i1> undef)
 ; ALL-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %V2 = call i1 @llvm.vector.reduce.and.v2i1(<2 x i1> undef)
 ; ALL-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %V4 = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> undef)
-; ALL-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %V6 = call i1 @llvm.vector.reduce.and.v6i1(<6 x i1> poison)
----------------
frederik-h wrote:

Sorry, I merged it incorrectly.

https://github.com/llvm/llvm-project/pull/208043