[llvm] [LV, VP]VP intrinsics support for the Loop Vectorizer + adding new tail-folding mode using EVL. (PR #76172)

[via llvm-commits]

================
@@ -5256,6 +5309,13 @@ LoopVectorizationCostModel::selectInterleaveCount(ElementCount VF,
   if (!isScalarEpilogueAllowed())
     return 1;
 
+  // Do not interleave if EVL is preferred and no User IC is specified.
+  if (foldTailWithEVL()) {
+    LLVM_DEBUG(dbgs() << "LV: Preference for VP intrinsics indicated. "
----------------
ayalz wrote:

Thanks for adding the message!
nit: is it more accurate to say that "LV: tail is folded with EVL, forcing unroll factor to be 1"?
The former could have been forced manually or preferred automatically, right?

https://github.com/llvm/llvm-project/pull/76172