[llvm] [AArch64] Improve lowering of truncating uzp1 (PR #82457)

[David Sherwood via llvm-commits]

================
@@ -21034,12 +21034,8 @@ static SDValue performUzpCombine(SDNode *N, SelectionDAG &DAG,
     }
   }
 
-  // uzp1(xtn x, xtn y) -> xtn(uzp1 (x, y))
-  // Only implemented on little-endian subtargets.
-  bool IsLittleEndian = DAG.getDataLayout().isLittleEndian();
-
-  // This optimization only works on little endian.
-  if (!IsLittleEndian)
+  // These optimizations only works on little endian.
----------------
david-arm wrote:

nit: These optimizations only work on little endian.

https://github.com/llvm/llvm-project/pull/82457