================
@@ -1578,11 +1581,92 @@ bool AMDGPUCodeGenPrepareImpl::visitLoadInst(LoadInst
&I) {
return false;
}
+// select (X == 0 || Y == 0), +0.0, (X * Y) -> fmul_legacy(X, Y)
+bool AMDGPUCodeGenPrepareImpl::foldSelectToFMulLegacy(SelectInst &I) const {
+ if (!I.getType()->isFloatTy())
+ return false;
+
+ // Inverted form: select (X != 0 && Y != 0), X * Y, +0.0.
+ bool Inverted = match(I.getFalseValue(), m_PosZeroFP());
+ if (!Inverted && !match(I.getTrueValue(), m_PosZeroFP()))
+ return false;
+
+ Value *X, *Y;
+ if (!match(Inverted ? I.getTrueValue() : I.getFalseValue(),
+ m_OneUse(m_FMul(m_Value(X), m_Value(Y)))))
+ return false;
+
+ // Collect the values compared against zero in the condition.
+ SmallVector<Value *, 4> CondWorklist = {I.getCondition()};
+ SmallVector<Value *, 4> CmpSrcs;
+ SmallPtrSet<Value *, 8> VisitedConds;
+ while (!CondWorklist.empty()) {
+ Value *Cond = CondWorklist.pop_back_val();
+ if (!VisitedConds.insert(Cond).second)
+ continue;
+
+ Value *LHS, *RHS, *Src;
+ if (Inverted ? match(Cond, m_And(m_Value(LHS), m_Value(RHS)))
+ : match(Cond, m_Or(m_Value(LHS), m_Value(RHS)))) {
+ CondWorklist.append({LHS, RHS});
+ continue;
+ }
+
+ if (!match(Cond, m_SpecificFCmp(Inverted ? FCmpInst::FCMP_UNE
+ : FCmpInst::FCMP_OEQ,
+ m_Value(Src), m_AnyZeroFP())))
+ return false;
+ CmpSrcs.push_back(Src);
+ }
+
+ SmallVector<Value *, 4> Roots[2] = {std::move(CmpSrcs), {X, Y}};
+ SmallPtrSet<Value *, 4> Leaves[2];
+ for (auto [Worklist, Found] : zip(Roots, Leaves)) {
+ SmallPtrSet<Value *, 8> Visited;
+
+ while (!Worklist.empty()) {
+ Value *Cur = Worklist.pop_back_val();
+ if (!Visited.insert(Cur).second)
+ continue;
+
+ // Peeling a nnan/ninf op could expose poison that the select hides.
+ auto *FPOp = dyn_cast<FPMathOperator>(Cur);
+ if (FPOp && (FPOp->hasNoNaNs() || FPOp->hasNoInfs())) {
+ Found.insert(Cur);
+ continue;
+ }
+
+ Value *Src;
+ // m_UnOp, not m_FNeg: fsub -0.0, X may flush a denormal result to zero.
+ if (match(Cur, m_CombineOr(m_UnOp(m_Value(Src)), m_FAbs(m_Value(Src)))))
{
+ Worklist.push_back(Src);
+ continue;
+ }
+
+ Found.insert(Cur);
+ }
+ }
+
+ if (Leaves[0] != Leaves[1])
+ return false;
+
+ IRBuilder<> Builder(&I);
+ Value *LegacyMul =
+ Builder.CreateIntrinsic(Intrinsic::amdgcn_fmul_legacy, {X, Y});
----------------
arsenm wrote:
Losing flags
https://github.com/llvm/llvm-project/pull/229357
_______________________________________________
llvm-branch-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
