[llvm] [DA] batch delinearization (PR #170519)

[Ryotaro Kasuga via llvm-commits]

================
@@ -58,21 +58,17 @@ for.end19:
   ret void
 }
 
+; With batch delinearization, the dependences are now computed correctly.
+; The interchange is still not profitable, but for a different reason.
----------------
kasuga-fj wrote:

I think this is unnecessary.

https://github.com/llvm/llvm-project/pull/170519