[llvm] [LV][AArch64] Provide option to use partial reductions by default (PR #216001)

[Sander de Smalen via llvm-commits]

================
@@ -36,13 +36,20 @@
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/Metadata.h"
 #include "llvm/Support/Casting.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/TypeSize.h"
 #include "llvm/Transforms/Utils/LoopUtils.h"
 
 using namespace llvm;
 using namespace VPlanPatternMatch;
 using namespace SCEVPatternMatch;
 
+static cl::opt<bool>
+    UsePartialReduceByDefault("use-partial-reduce-by-default", cl::init(false),
+                              cl::Hidden,
+                              cl::desc("Use partial reduction intrinsics for "
+                                       "all supported unordered reductions."));
+
----------------
sdesmalen-arm wrote:

Should the option contain a TODO to explain that we plan to remove this as soon as we have confidence that partial.reduce.add has no worse handling as vector add?

https://github.com/llvm/llvm-project/pull/216001