[llvm] [LV] Prefer fixed VFs for epilogue vectorization (PR #211814)

[Benjamin Maxwell via llvm-commits]

================
@@ -706,6 +706,21 @@ bool LoopVectorizationPlanner::isMoreProfitable(const VectorizationFactor &A,
         !B.Width.isScalar())
       return true;
 
+  // Prefer fixed VFs for epilogue loops (unless the fixed cost is much more
+  // expensive -- >= 2x). There are some extra costs for using scalable vectors
----------------
MacDue wrote:

Ah, I didn't realize InstructionCost used a fixed point arith (I assumed integer division that'd truncate towards zero). 

https://github.com/llvm/llvm-project/pull/211814