[llvm] [LoopVectorize][AArch64][SVE] Generate wide active lane masks (PR #81140)

Florian Hahn via llvm-commits llvm-commits at lists.llvm.org
Thu Jun 6 02:55:03 PDT 2024


================
@@ -1329,6 +1329,50 @@ class VPInstruction : public VPRecipeWithIRFlags {
   }
 };
 
+class VPActiveLaneMaskRecipe : public VPRecipeWithIRFlags {
----------------
fhahn wrote:

At the moment, it's not clear why it is needed to move this out of VPInstruction, as it still only uses an opcode + operands and no extra data. Depending on what information exactly is used for widen codegen, there may be the need to have a separate class, but in general using VPInstruction when the information can be encoded easily via opcode + VPValue operands only is preferred.

https://github.com/llvm/llvm-project/pull/81140


More information about the llvm-commits mailing list