Author: Lance Perkins
Date: 16:50:13 07/13/04
Go up one level in this thread
Yeah, and that few percent is all it takes for Thinker to loose. Btw, there are already countless tournaments like that. I wouldn't bother going half-way around the world for that. I can do that tournament myself. And if I did join, I will later complain about my non-GM-made book and my cheap-old VC5 compiler, my small eval, the lack of EGTB, non-optimal move-ordering, weaker king safety, and a host of other "non-equal" things. In short, no, thanks. Its not the WCCC that developers will go to. Btw, if I could get parallel search working on Thinker, I 'might' join next year's WCCC, and bring whatever is the latest HW I can afford by that time. --- On July 13, 2004 at 19:25:58, Omid David Tabibi wrote: >On July 13, 2004 at 19:21:45, Lance Perkins wrote: > >>So if engine A and engine B both run on the same AMD hardware, then the playing >>field is equal? Sorry to disagree here. If engine A is optimized for AMD and >>engine B is optmized for Intel, it obviously not equal. Thinker, for example, is >>optmized for Intel only. Would I optmized it for AMD just to join your AMD-only >>tournament? I would not. Its not fair. > >It would be safe to assume that the difference between the "optimized for AMD" >and non-optimized versions wouldn't be 400%... and in fact in most cases it >would be only a few percent. > >> >> >>On July 13, 2004 at 19:04:39, Omid David Tabibi wrote: >> >>>On July 13, 2004 at 18:56:59, Lance Perkins wrote: >>> >>>>So, after all that ranting about having the "same" machine, to you, the >>>>difference in processor architecture (AMD vs Intel) does not matter. >>>> >>>>Well, it does. So, is the AMD 3200+ (2Ghz), the same as 3.2Gz Intel? >>> >>>No it is not. But if all participants get the same thing it won't matter much. >>> >>>>Or should >>>>it be a 2Ghz AMD vs 2Ghz Intel. It really gets muddy. >>> >>>It doesn't matter whether it is Intel or AMD, 1GHz or 2GHz, as long as it is >>>equal for everyone. >>> >>> >>>> >>>>--- >>>> >>>>On July 13, 2004 at 18:42:52, Omid David Tabibi wrote: >>>> >>>>>On July 13, 2004 at 18:37:18, Lance Perkins wrote: >>>>> >>>>>>Using your horse race analogy: after a horse race, are you comparing horses or >>>>>>are you comparing jockeys? >>>>> >>>>>Both of them, assuming all participants used horses, and not donkies. >>>>> >>>>> >>>>>> >>>>>>Now back to chess engines... Even on the same hardware, comparison is not >>>>>>accurate. What if a program is optimized for Intel architecture, while another >>>>>>is designed for AMD architecture? Picking one hardware would be unfair for one >>>>>>of the engines. >>>>> >>>>>Most chess programs run better on AMD. But it won't matter if all of them run on >>>>>Intel or all of them run on AMD (at most it will affect them by a few percentage >>>>>in comparison to other programs, surely not 400%...). >>>>> >>>>> >>>>>> >>>>>>So really, what is your definition of the "world 'computer' chess 'champion'"? >>>>>> >>>>>>On July 13, 2004 at 18:24:50, Omid David Tabibi wrote: >>>>>> >>>>>>>On July 13, 2004 at 18:15:42, Lance Perkins wrote: >>>>>>> >>>>>>>>Are you saying that given the same hardware, games between chess programs are >>>>>>>>"fair" by your definition? >>>>>>>> >>>>>>>>How can it be if they use different opening books? >>>>>>>> >>>>>>>>What people put in the code is not even the same. >>>>>>> >>>>>>>It depends what you want to compare. If you want to compare chess playing >>>>>>>strength, then of course opening books, evaluation, etc, are all part of the >>>>>>>comparison. If you want to compare hardware, then run a unique program >>>>>>>(benchmark) on each hardware and compare the results. >>>>>>> >>>>>>>But by the current format of WCCC you don't reach any of these conclusions. >>>>>>> >>>>>>> >>>>>>>> >>>>>>>>Arasan's eval code has 3700 lines of code. Thinker has less than 1000 LOC. >>>>>>>>Arasan's binary is 1.3M (1300K). Thinker is 80K (less that 16 times). Have you >>>>>>>>seen a boxing match between 50lb kid and a 180lb adult? >>>>>>>> >>>>>>>>Maybe folks are having difficulty with the idea that a chess program is a >>>>>>>>combination of hardware and software. Can you really do anything with just >>>>>>>>sotware? >>>>>>>> >>>>>>>>The world "computer" chess champion is the best H+S combo. >>>>>>>> >>>>>>>>Cheers... >>>>>>>> >>>>>>>>On July 13, 2004 at 17:45:30, Omid David Tabibi wrote: >>>>>>>> >>>>>>>>>On July 13, 2004 at 17:40:19, Lance Perkins wrote: >>>>>>>>> >>>>>>>>>>In a "war", that's how it is. >>>>>>>>> >>>>>>>>>Nobody claimed war is fair. >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>>On July 13, 2004 at 17:38:42, Omid David Tabibi wrote: >>>>>>>>>> >>>>>>>>>>>The 2004 World Boxing Championship >>>>>>>>>>> >>>>>>>>>>>Rules: >>>>>>>>>>>Each participant is allowed to use whatever weaponry he would like. >>>>>>>>>>> >>>>>>>>>>>Participants: >>>>>>>>>>> >>>>>>>>>>>Name Weapon >>>>>>>>>>>---- ------ >>>>>>>>>>>Bill no weapon >>>>>>>>>>>Jack knife >>>>>>>>>>>Jonathan handgun >>>>>>>>>>>George M16 rifle >>>>>>>>>>>Robert RPG >>>>>>>>>>> >>>>>>>>>>>May the best boxer win!
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.