[llvm] [LLVM][SVE] Implement isel for bfloat fptoi and itofp operations. (PR #129713)

[David Sherwood via llvm-commits]

================
@@ -4733,6 +4753,10 @@ SDValue AArch64TargetLowering::LowerVectorFP_TO_INT(SDValue Op,
       return DAG.getSetCC(DL, VT, Cvt, Zero, ISD::SETNE);
     }
 
+    // Let common code split the operation.
----------------
david-arm wrote:

I assume the reason you've put this after the i1 case above is because you believe that when converting nxv8bf16 -> nxv8i1 it's better to do:

FP_EXTEND: nvx8bf16 -> nxv8f32
VectorFP_TO_INT: nxv8f32 -> nxv8i32
SETNE: nxv8f32, zero -> nxv8i1

than

FP_EXTEND: nvx8bf16 -> nxv8f32
VectorFP_TO_INT: nxv8f32 -> nxv8i1

Presumably because you think SETNE will do a better job of splitting with an i1 result element type, than VectorFP_TO_INT?

https://github.com/llvm/llvm-project/pull/129713