This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RFC] Run copyrename one extra time before out-of-ssa
On Sunday 23 January 2005 02:42, Steven Bosscher wrote:
> On Friday 21 January 2005 18:59, Mark Mitchell wrote:
> > Jeffrey A Law wrote:
> > >>May I interpret your comments as "OK" after bootstrapping/testing? ;-)
> > >
> > > I can't speak for Andrew, but it would have my support.
> >
> > In theory, this makes sense, but we need more analysis than an isolated
> > testcase. Please obtain benchmark numbers on at least one, and
> > hopefully two or three, platforms. If this is not a demonstrable win,
> > then let's not do it.
>
> OK, so that went a lot smoother than I expected...
>
> Bootstrapped and tested on i686, ia64, x86_64, ppc64, and ppc. There
> are no new regressions, of course.
>
> Benchmark results are not all in yet, but for AMD64 I have SPECint and
> SPECfp numbers, at -O2. Here is a summary of the results:
>
> - Not one binary was bigger, and a few were smaller with the patch.
> Overall the binaries are ~0.5% smaller.
> - Compile time was 2s faster on 194 seconds for the unpatched compiler.
> Let's call it compile time neutral.
> - For SPECint there is a 5% win for twolf, all other benchmarks are
> within 0.5%.
> - For SPECfp there is a 30% improvement for mgrid (!), and 6% for mesa.
> The rest is in the noise (within 0.4%).
The results for i686 are more mixed, but overall still better with the
patch (peak) than without. gap failed to build and vortex miscompared,
I'm not sure why. twolf, eon, and wupwise take a small hit, but perlbmk
and sixtrack soar. Compile is unchanged. The binaries are smaller
again overall, and no individual binaries are bigger, except for eon
(529442 bytes base (ie. unpatched) and 529602 bytes peak).
SPECint:
164.gzip 1400 201 697* 1400 199 702* +
175.vpr 1400 206 681* 1400 203 689* +
176.gcc 1100 125 883* 1100 125 881* -
181.mcf 1800 338 533* 1800 339 532* -
186.crafty 1000 109 921* 1000 108 924* +
197.parser 1800 288 626* 1800 285 632* +
252.eon 1300 135 961* 1300 139 938* -
253.perlbmk 1800 216 834* 1800 193 933* +
254.gap X X ?
255.vortex X X ?
256.bzip2 1500 222 677* 1500 221 679* +
300.twolf 3000 303 992* 3000 310 968* -
Est. SPECint_base2000 766
Est. SPECint2000 773
SPECfp:
168.wupwise 1600 255 628 * 1600 265 603 * -
171.swim 3100 447 693 * 3100 447 693 * =
172.mgrid 1800 489 368 * 1800 487 370 * +
173.applu 2100 364 577 * 2100 362 580 * +
177.mesa 1400 206 681 * 1400 203 690 * +
178.galgel X X ?
179.art 2600 541 480 * 2600 536 485 * +
183.equake 1300 159 818 * 1300 158 821 * +
187.facerec 1900 309 615 * 1900 309 614 * -
188.ammp 2200 311 708 * 2200 310 710 * +
189.lucas 2000 288 694 * 2000 289 693 * -
191.fma3d 2100 301 697 * 2100 302 695 * -
200.sixtrack 1100 314 350 * 1100 265 415 * +
301.apsi 2600 369 705 * 2600 380 684 * -
Est. SPECfp_base2000 599
Est. SPECfp2000 605
My ppc run was ruined by something running in the back ground, so I do
not have numbers for it :-/
Gr.
Steven