[LLVMdev] Performance regression on ARM

[Chandler Carruth]

On Thu, Oct 16, 2014 at 9:56 PM, Adam Nemet <anemet at apple.com> wrote:

> > On Oct 16, 2014, at 3:04 AM, Renato Golin <renato.golin at linaro.org>
> wrote:
> >
> > On 16 October 2014 09:34, Hal Finkel <hfinkel at anl.gov> wrote:
> >> Interesting. Looks like the problem is in (219545, 219569].
> >
> > Yes.
>
> Chandler’s complex arithmetic changes are also in the range: r219557 in
> clang.  We saw it change the code in mandel-2 significantly.


I would not at all be surprised by regression in some cases here.

Complex arithmetic with reals got *much* faster. Complex arithmetic with
other complex numbers may have gotten slower. If there were fastmath flags,
*crazy preposterously slower*.

Note that I have a patch out needing review which addresses most of the
regression in this case, and should address all of the regressions if
fastmath flags are in play.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20141016/3dceba98/attachment.html>