[llvm] [InstCombine][RISCV] Convert VPIntrinsics with splat operands to splats (PR #65706)

[Luke Lau via llvm-commits]

================
@@ -729,6 +730,96 @@ bool VectorCombine::foldBitcastShuf(Instruction &I) {
   return true;
 }
 
+/// VP Intrinsics whose vector operands are both splat values may be simplified
+/// into the scalar version of the operation and the result is splatted. This
+/// can lead to scalarization down the line.
+bool VectorCombine::scalarizeVPIntrinsic(VPIntrinsic &VPI) {
+  Value *Op0 = VPI.getArgOperand(0);
+  Value *Op1 = VPI.getArgOperand(1);
+
+  if (!getSplatValue(Op0) || !getSplatValue(Op1))
+    return false;
+
+  // For the binary VP intrinsics supported here, the result on disabled lanes
+  // is a poison value. For now, only do this simplification if all lanes
+  // are active.
+  // TODO: Relax the condition that all lanes are active by using insertelement
+  // on inactive lanes.
+  auto IsAllTrueMask = [](Value *MaskVal) {
+    if (Value *SplattedVal = getSplatValue(MaskVal))
+      if (auto *ConstValue = dyn_cast<Constant>(SplattedVal))
+        return ConstValue->isAllOnesValue();
+    return false;
+  };
+  if (!IsAllTrueMask(VPI.getArgOperand(2)))
+    return false;
+
+  // Check to make sure we support scalarization of the intrinsic
+  std::set<Intrinsic::ID> SupportedIntrinsics(
+      {Intrinsic::vp_add, Intrinsic::vp_sub, Intrinsic::vp_mul,
+       Intrinsic::vp_ashr, Intrinsic::vp_lshr, Intrinsic::vp_shl,
+       Intrinsic::vp_or, Intrinsic::vp_and, Intrinsic::vp_xor,
+       Intrinsic::vp_fadd, Intrinsic::vp_fsub, Intrinsic::vp_fmul,
+       Intrinsic::vp_sdiv, Intrinsic::vp_udiv, Intrinsic::vp_srem,
+       Intrinsic::vp_urem});
+  Intrinsic::ID IntrID = VPI.getIntrinsicID();
+  if (!SupportedIntrinsics.count(IntrID))
+    return false;
+
+  // Calculate cost of splatting both operands into vectors and the vector
+  // intrinsic
+  VectorType *VecTy = cast<VectorType>(VPI.getType());
+  TTI::TargetCostKind CostKind = TTI::TCK_RecipThroughput;
+  InstructionCost SplatCost =
+      TTI.getVectorInstrCost(Instruction::InsertElement, VecTy, CostKind, 0) +
+      TTI.getShuffleCost(TargetTransformInfo::SK_Broadcast, VecTy);
+
+  // Calculate the cost of the VP Intrinsic
+  SmallVector<Type *, 4> Args;
+  for (Value *V : VPI.args())
+    Args.push_back(V->getType());
+  IntrinsicCostAttributes Attrs(IntrID, VecTy, Args);
+  InstructionCost VectorOpCost = TTI.getIntrinsicInstrCost(Attrs, CostKind);
+  InstructionCost OldCost = 2 * SplatCost + VectorOpCost;
+
+  // Calculate cost of scalarizing
+  std::optional<unsigned> ScalarOpcodeOpt =
+      VPIntrinsic::getFunctionalOpcodeForVP(IntrID);
+  assert(ScalarOpcodeOpt && "Unable to determine scalar opcode");
+  unsigned ScalarOpcode = *ScalarOpcodeOpt;
+
+  InstructionCost ScalarOpCost =
+      TTI.getArithmeticInstrCost(ScalarOpcode, VecTy->getScalarType());
+  InstructionCost NewCost = ScalarOpCost + SplatCost;
----------------
lukel97 wrote:

The existing splats might still be kept around by other instructions, should we check if they have any other users and incorporate that into to the new cost?

https://github.com/llvm/llvm-project/pull/65706