Author: Vincent Diepeveen
Date: 05:12:49 05/27/02
Go up one level in this thread
On May 27, 2002 at 08:07:44, Vincent Diepeveen wrote: note that IF you manage to get it to work with intel c++ 5.0, you first want to generate profile information, then after that recopmile to get it faster using that profile information for branches. intel c++ 5.0 CFLAGS = -O3 -G6 -Qaxi -Qxi -Gr -Qprof_genx then run executable for 10 minutes. then recompile with: CFLAGS = -O3 -G6 -Qaxi -Qxi -Gr -Qprof_use apart from the -D defined from crafty which you need to add to this. this produces a P3 optimized executable. the P4 optimizations at intel are very buggy. should probably be seen as a -specbench optimization :) >On May 27, 2002 at 08:00:35, Slater Wold wrote: > >INTEL on an AMD processor. Please let me quote the official >explanation from INTEL C++ team: > > "We did of course not test on the K7 processor, so whether > it is slower or faster on the K7 processor than on the intel > processor we do not know". > >I'll throw you an exe. Take into account that the linux equivalent >is another 5% to 12.5% faster than that one is for the K7. > >Bob isn't awake yet. he'll get soon. > >Crafty as usual has an error when compiling under intel. > >Your first posting here said you used the default compile from bob. >how the HELL can you compile for intel 5.0 if the default 1815 source >doesn't compile at all here? > >>Like I said Vince, throw me any EXE you want, and I guarantee the plain 'ol exe >>off Hyatt's site will beat any Crafty EXE you can give. >> >>I used the beloved Intel 5.0 to build an exe for the AMD, and it was still >>slower. And everyone knows, you CANNOT get a faster exe than that. >> >>On May 27, 2002 at 07:43:31, Vincent Diepeveen wrote: >> >>>On May 27, 2002 at 07:19:34, Slater Wold wrote: >>> >>>>On May 27, 2002 at 07:07:21, Jorge Pichard wrote: >>>> >>>>>If you auto test then in a short match of 20 games it could go either way, since >>>>>for chess, the speed difference is NOT noticeable. >>>>> >>>>>Jorge. >>>> >>>>Chess is not all I do. Nor is it all anyone does on their PC. >>> >>>>100% (literally) better graphics than the AMD, using the same video card. >>> >>>bandwidth is bigger on P4 sure. >>> >>>>30% faster encoding video. 15% faster encoding MP3's. >>> >>>same as above. this is all bandwidth limited. >>> >>>>These are *not* marginal victories. These are HUGE gains. >>>> >>>>I said in my original post, that I was interested in chess performance, but >>>>let's be honest here for a second, most people who have a PC don't have a chess >>>>program installed. >>>> >>>>Bottom line: Whatever you do with your Intel P4 2.53Ghz machine, it will be >>>>faster than my AMD 1.73Ghz machine. If that's by 10% or 100%, it's going to be >>>>faster. >>> >>>Not at all. Note that gcc also works under windows (they say), >>>but the important thing is that there are at least 2 compilers for >>>windows from which 1 is a very important compiler the visual c++ compiler, >>>which is way faster for the K7 than intel is. >>> >>>You got fooled again by the intel compiler. >>> >>>Please look at specint and you'll see that most tests they use intel c++ >>>compiled executables. >>> >>>It's sick.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.