Serious performance regression -- some tree optimizer questions

[Andrew Pinski]

On Dec 17, 2004, at 4:24 PM, Ulrich Weigand wrote:

> Hello,
>
> we've started to do some performance analysis of current GCC mainline
> on zSeries, and while in general results are promising, we've found a
> couple of serious performance regressions.  The most extreme one was
> the SPECfp2000 test case mgrid -- on s390x optimized for z990, this
> test case takes about *twice* the run time with current mainline as
> it did with GCC 3.4 ...
>
> However, current mainline does rather badly at all three of
> these tasks.  I'm not quite sure which pass is at fault here;
> there appear to be some optimizations that would appear quite
> straightforward to me that no pass is currently performing.
>
> Any suggestions how to further investigate this?  Is this just
> the way things are with the current tree optimizers, or is this
> supposed to work and we just need to find the bug?

I should note that there were a recent patch to IVOPTs that was
approved but not yet applied (why, I don't know) that should help.
This patch helps us find the global minimum instead of a local
minimum.  Hopefully this patch helps this case also, it might not
but who knows.

-- Pinski