[Mlir-commits] [mlir] Fix TOSA FP16->INT16 CAST lowering (PR #79299)

[Thomas Preud'homme]

================
@@ -480,23 +480,53 @@ createLinalgBodyCalculationForElementwiseOp(Operation *op, ValueRange args,
     }
 
     if (arith::FPToSIOp::areCastCompatible(srcTy, dstTy)) {
-      auto intMin = rewriter.create<arith::ConstantOp>(
+      auto intMinFP = rewriter.create<arith::ConstantOp>(
           loc, rewriter.getFloatAttr(
                    getElementTypeOrSelf(srcTy),
                    APInt::getSignedMinValue(dstTy.getIntOrFloatBitWidth())
                        .getSExtValue()));
 
-      auto intMax = rewriter.create<arith::ConstantOp>(
+      auto rounded = rewriter.create<math::RoundEvenOp>(loc, args[0]);
+
+      // The input floating-point type has enough mantissa bits to represent
+      // the max int value so just clamp the input in the floating-point
+      // domain and convert to int. Note: the min value can be represented
+      // because it consists of a mantissa with only the lsb set.
+      if (cast<FloatType>(srcTy).getFPMantissaWidth() >=
+          dstTy.getIntOrFloatBitWidth() - 1) {
----------------
RoboTux wrote:

Actually for FP16->I32 we need to clamp -inf and +inf. I've addressed this in the updated patch.

https://github.com/llvm/llvm-project/pull/79299