This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] Enable fmove-loop-invariants by default
Hello,
> 1. Test the performance impact of removing loop.c with SPEC on popular
> platforms, like, say, AMD64, ARM, PowerPC, and Itanium. (A somewhat
> random set, certainly prejudiced by my own experience; the goal is to
> get a mix of workstation and embeded processors.)
results from ia64-linux (overall neutral for specfp, 0.6% regression on
specint).
Zdenek
168.wupwise 1600 342 468 * 1600 347 461 *
171.swim 3100 574 540 * 3100 578 537 *
172.mgrid 1800 579 311 * 1800 579 311 *
173.applu 2100 543 387 * 2100 542 387 *
177.mesa 1400 181 772 * 1400 177 790 *
178.galgel X X
179.art 2600 129 2022 * 2600 127 2041 *
183.equake 1300 384 339 * 1300 383 339 *
187.facerec 1900 466 408 * 1900 461 412 *
188.ammp 2200 309 711 * 2200 312 704 *
189.lucas 2000 254 787 * 2000 254 787 *
191.fma3d 2100 828 254 * 2100 821 256 *
200.sixtrack 1100 345 318 * 1100 353 312 *
301.apsi 2600 512 507 * 2600 512 508 *
Est. SPECfp_base2000 508
Est. SPECfp2000 508
164.gzip 1400 197 710* 1400 198 707*
175.vpr 1400 159 882* 1400 161 872*
176.gcc 1100 93.9 1172* 1100 94.2 1168*
181.mcf 1800 186 967* 1800 194 926*
186.crafty 1000 108 922* 1000 109 919*
197.parser 1800 253 712* 1800 252 713*
252.eon 1300 171 759* 1300 170 763*
253.perlbmk X X
254.gap 1100 213 515* 1100 216 510*
255.vortex X X
256.bzip2 1500 203 740* 1500 203 741*
300.twolf 3000 286 1050* 3000 284 1056*
Est. SPECint_base2000 822
Est. SPECint2000 817