[PATCH] D81267: [LV] Enable the LoopVectorizer to create pointer inductions

[Florian Hahn via Phabricator via llvm-commits]

fhahn added inline comments.


================
Comment at: llvm/test/Transforms/LoopVectorize/pointer-induction.ll:4
+target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
+target triple = "x86_64-unknown-linux-gnu"
+
----------------
dmgreen wrote:
> I think if you use x86 as a target (and needs it for the costing), the test needs to go into test/Transforms/LoopVectorize/X86 in case the target is not compiled in.
It looks like the options above actually force vectorization with a certain factor. In that case, it Is probably best to remove the triple.

I'd also consider just checking the loop-vectorize output (without -dce -instcombine), if it is not too messy, as it makes the test more prone to break when something changes in instcombine. Also, it might be possible to only specifically check the IR related to the generated induction, rather than autogenerating the checks, which include a lot of relatively irrelevant stuff.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D81267/new/

https://reviews.llvm.org/D81267