[llvm] [SLP][AMDGPU] Vectorize operands of non-trivially-vectorizable intrinsic calls (PR #189784)

[Alexey Bataev via llvm-commits]

================
@@ -18445,10 +18476,9 @@ InstructionCost BoUpSLP::getTreeCost(InstructionCost TreeCost,
           assert(SLPReVec && "Only supported by REVEC.");
           SrcTy = getWidenedType(SrcTy, VecTy->getNumElements());
         }
-        InstructionCost CastCost =
-            TTI->getCastInstrCost(Opcode, DstTy, SrcTy,
-                                  TTI::CastContextHint::None,
-                                  TTI::TCK_RecipThroughput);
+        InstructionCost CastCost = TTI->getCastInstrCost(
----------------
alexey-bataev wrote:

Restore original formatting

https://github.com/llvm/llvm-project/pull/189784