[llvm] [AArch64][NEON] Lower fixed-width add partial reductions to dot product (PR #107078)

[David Sherwood via llvm-commits]

================
@@ -21781,7 +21782,8 @@ SDValue tryLowerPartialReductionToDot(SDNode *N,
              Intrinsic::experimental_vector_partial_reduce_add &&
          "Expected a partial reduction node");
 
-  if (!Subtarget->isSVEorStreamingSVEAvailable())
+  if (!Subtarget->isSVEorStreamingSVEAvailable() &&
----------------
david-arm wrote:

What if this is <4 x i32> and we don't have NEON's dot product, but we do have SVE? Will we do something sensible? If we lower fixed-width using SVE, then we should add an extra RUN line to the tests for this.

https://github.com/llvm/llvm-project/pull/107078