SLP/Loop vectorizer pass ordering

[James Molloy]

I could do that - It'd be a simple patch. I'll commit it later today.


On 6 August 2014 10:42, Chandler Carruth <chandlerc at google.com> wrote:

>
> On Wed, Aug 6, 2014 at 2:37 AM, James Molloy <james at jamesmolloy.co.uk>
> wrote:
>
>> Hi Hal,
>>
>> I ran some numbers on Cortex-A57 (AArch64 mode) with the SLP and Loop
>> vectorizers swapped and with your noalias metadata enabled. I see no change
>> outside of noise. I ran SPEC(2000+2006) and EEMBC - LNT is currently
>> unusable on our hardware due to lack of "perf" - it's just too noisy :(
>>
>> Can someone try a test on x86? If that shows the same, we should be in a
>> good position to swap the ordering as soon as you enable noalias metadata
>> by default.
>>
>
> Can you set everything up so that there is a flag that makes this switch?
> Or is there already one I missed?
>
> Once I have flags, its easy for me to test. I can test both just the AA
> metadata, and then AA metadata + vectorize rearrangement.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140806/d393a508/attachment.html>