[llvm] [AMDGPU] Select (xor i1 (divergent trunc:i32 x), -1) -> cmp_neq x, 1 (PR #133698)

[Ana Mihajlovic via llvm-commits]

================
@@ -1267,6 +1267,19 @@ class DivergentUnaryFrag<SDPatternOperator Op> : PatFrag <
   let GISelPredicateCode = [{return true;}];
 }
 
+class DivergentUnaryFrag_oneuse<SDPatternOperator Op> : PatFrag <
----------------
mihajlovicana wrote:

seems to be working

https://github.com/llvm/llvm-project/pull/133698