[llvm-branch-commits] [llvm] [AMDGPU] Treat a bitcast between legal types or same-size scalars as free (PR #238412)

Matt Arsenault via llvm-branch-commits llvm-branch-commits at lists.llvm.org
Thu Nov 5 17:02:19 PST 2026


================
@@ -1066,6 +1066,19 @@ InstructionCost GCNTTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst,
         Elts && Src->isIntegerTy(*Elts))
       return InstructionCost(MaskUnpackCostPerElt) * *Elts *
              getFullRateInstrCost();
+    // The generic code scalarizes a vector cast into element casts, which for
+    // a bitcast may pair elements of different sizes.
+    if (Src->getPrimitiveSizeInBits() == Dst->getPrimitiveSizeInBits()) {
+      // Integer and FP values share registers. A legalized vector may need its
+      // lanes repacked.
+      if (TLI->isTypeLegal(TLI->getValueType(DL, Src)) &&
+          TLI->isTypeLegal(TLI->getValueType(DL, Dst)))
+        return 0;
+      if (!Src->isVectorTy() && !Dst->isVectorTy() &&
+          getTypeLegalizationCost(Src).second.getSizeInBits() ==
+              getTypeLegalizationCost(Dst).second.getSizeInBits())
+        return 0;
----------------
arsenm wrote:

Pretty sure this ends up doing the same Type -> EVT calculation multiple times 

https://github.com/llvm/llvm-project/pull/238412


More information about the llvm-branch-commits mailing list