[flang-commits] [flang] [flang] Use saturated intrinsic for floating point conversions (PR #130686)

[Kiran Chandramohan via flang-commits]

================
@@ -835,10 +835,20 @@ struct ConvertOpConversion : public fir::FIROpConversion<fir::ConvertOp> {
         return mlir::success();
       }
       if (mlir::isa<mlir::IntegerType>(toTy)) {
-        if (toTy.isUnsignedInteger())
-          rewriter.replaceOpWithNewOp<mlir::LLVM::FPToUIOp>(convert, toTy, op0);
-        else
-          rewriter.replaceOpWithNewOp<mlir::LLVM::FPToSIOp>(convert, toTy, op0);
+        // NOTE: We are checking the fir type here because toTy is an LLVM type
+        // which is signless, and we need to use the intrinsic that matches the
+        // sign of the output in fir.
+        if (toFirTy.isUnsignedInteger()) {
+          auto intrinsicName =
+              mlir::StringAttr::get(convert.getContext(), "llvm.fptoui.sat");
+          rewriter.replaceOpWithNewOp<mlir::LLVM::CallIntrinsicOp>(
+              convert, toTy, intrinsicName, op0);
+        } else {
+          auto intrinsicName =
+              mlir::StringAttr::get(convert.getContext(), "llvm.fptosi.sat");
+          rewriter.replaceOpWithNewOp<mlir::LLVM::CallIntrinsicOp>(
+              convert, toTy, intrinsicName, op0);
+        }
----------------
kiranchandramohan wrote:

Would you know what these saturation intrinsics get lowered to? And is there a big difference in performance? Would it be possible to use the saturation intrinsic only when necessary? Or can that not be determined at compile time?

https://github.com/llvm/llvm-project/pull/130686