[Mlir-commits] [mlir] [mlir][ArmSVE] Add intrinsics for the SME2 multi-vector zips (PR #80985)

[Benjamin Maxwell]

================
@@ -410,4 +411,24 @@ def ConvertToSvboolIntrOp :
     /*overloadedResults=*/[]>,
     Arguments<(ins SVEPredicate:$mask)>;
 
+// Note: This multi-vector intrinsic requires SME2.
+def ZipX2IntrOp : ArmSVE_IntrOp<"zip.x2",
+    /*traits=*/[],
----------------
MacDue wrote:

All the SVE intrinsics are super lax like this. '`AnyScalableVector`' is much more than these can actually take, I've left the type checking to the actual ops. 

https://github.com/llvm/llvm-project/pull/80985