[cairo] Re: _cairo_lround

[Daniel Amelang]

On 11/22/06, Koen Kooi <koen at dominion.kabel.utwente.nl> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Koen Kooi schreef:
> > Daniel Amelang schreef:
> >> On 11/21/06, Koen Kooi <koen at dominion.kabel.utwente.nl> wrote:
> >>> -----BEGIN PGP SIGNED MESSAGE-----
> >>> Hash: SHA1
> >>>
> >>> Daniel Amelang schreef:
> >>>
> >>>> This doesn't help _that_ much on systems with an FPU (although it
> >>>> doesn't hurt), but it does provide a 1.5x speedup for the text cases
> >>>> on my softfloat tests. Notice that I'm using a x86 softfloat library
> >>>> (with a libm also compiled for softfloat) running on my x86 machine,
> >>>> so someone with a Nokia 770 needs to post their results for us to get
> >>>> something truly meaningful.
> >>> Does it have to be a nokia770? I have lots arm toys running X, but my
> >>> n770 is needed for
> >>> day to day uses.
> >> Uh, not really. I think any ARM that doesn't have a FPU should produce
> >> similiar numbers. Unless the X server version makes a diff, I suppose.
> >> Still, the numbers would be useful.
> >
> > The perf for the vanilla git can be seen at http://www.angstrom-distribution.org/~koen/cairo/
> >
> > I'll make some .perfs with the patches later on and upload them to the same location,
> > including the output(s) of cairo-perf-diff.
>
> Or not:
>
> root at h2200:/data/cairo/perf# gdb .libs/cairo-perf-diff-files
> GNU gdb 6.4
> Copyright 2005 Free Software Foundation, Inc.
> GDB is free software, covered by the GNU General Public License, and you are
> welcome to change it and/or distribute copies of it under certain conditions.
> Type "show copying" to see the conditions.
> There is absolutely no warranty for GDB.  Type "show warranty" for details.
> This GDB was configured as "arm-angstrom-linux-gnueabi"...Using host libthread_db library
> "/lib/libthread_db.so.1".
>
> (gdb) run git-200610220-vanilla.perf foo.perf
> Starting program: /data/cairo/perf/.libs/cairo-perf-diff-files git-200610220-vanilla.perf
> foo.perf
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x00009d78 in _cairo_stats_compute ()
> (gdb)
>
> Cairo was natively compiled on the target.
>
> any hints on this?

Yes, I think cairo-diff-files is broken in git at the moment, as I
also get a segfault on my x86. Actually, it's been broken for me for a
week now, I just haven't gotten around to reporting it to Carl.

If you just put up the .perf files, that's all I need, as I have a
ruby script that does a simplified version of cairo-perf-diff-files.
Looks like the vanilla is already up, I just need the patched results.

Thanks again!