[llvm] [GlobalISel] Add support for interleave and deinterleave intrinsics to IRTranslator (PR #85199)

[Dhruv Chawla via llvm-commits]

================
@@ -2474,6 +2475,49 @@ bool IRTranslator::translateKnownIntrinsic(const CallInst &CI, Intrinsic::ID ID,
 
     return true;
   }
+
+  case Intrinsic::experimental_vector_interleave2: {
+    Value *Src0 = CI.getOperand(0);
+    Value *Src1 = CI.getOperand(1);
+
+    // Canonicalize fixed-width vector types to G_SHUFFLE_VECTOR
+    // (similar to SelectionDAG)
+    LLT OpType = getLLTForType(*Src0->getType(), MIRBuilder.getDataLayout());
+    if (!OpType.isFixedVector())
+      break;
+
+    Register Op0 = getOrCreateVReg(*Src0);
+    Register Op1 = getOrCreateVReg(*Src1);
+    Register Res = getOrCreateVReg(CI);
+
+    MIRBuilder.buildShuffleVector(
+        Res, Op0, Op1, createInterleaveMask(OpType.getNumElements(), 2));
+
+    return true;
----------------
dc03-work wrote:

> call based on the condition to be met.

I'm not entirely sure what you mean by this, can you please clarify?

https://github.com/llvm/llvm-project/pull/85199