[llvm] [NVPTX] Implemented istruncatefree and iszextfree alongwith test cases. (PR #115139)

[Justin Fargnoli via llvm-commits]

================
@@ -506,6 +506,17 @@ class NVPTXTargetLowering : public TargetLowering {
            DstTy->getPrimitiveSizeInBits() == 32;
   }
 
+  bool isTruncateFree(EVT FromVT, EVT ToVT) const override {
+    if (!FromVT.isScalarInteger() || !ToVT.isScalarInteger()) {
+      return false;
+    }
+    return FromVT.getSizeInBits() == 64 && ToVT.getSizeInBits() == 32;
+  }
+
+  bool isZExtFree(EVT FromVT, EVT ToVT) const override { return false; }
+
+  bool isZExtFree(Type *SrcTy, Type *DstTy) const override { return false; }
----------------
justinfargnoli wrote:

I've asked around internally and have concluded that although `zext` may be cheap or possible to optimize away, it is *not* free in the general case. 

So, the implementation you went with is correct. However, it isn't buying us anything because the [default behavior](https://github.com/llvm/llvm-project/blob/4d4024e1edf354113e8d0d11661d466ae5b0bee7/llvm/include/llvm/CodeGen/TargetLowering.h#L3046-L3060) is to return false.  

Thus, let's delete the overrides and narrow the scope of this PR! 

https://github.com/llvm/llvm-project/pull/115139