Author: Vincent Diepeveen
Date: 14:46:06 09/26/03
Go up one level in this thread
On September 26, 2003 at 16:49:22, Christophe Theron wrote: >On September 25, 2003 at 23:02:20, Vincent Diepeveen wrote: > >>On September 25, 2003 at 12:53:42, Christophe Theron wrote: >> >>>On September 25, 2003 at 09:41:05, Vincent Diepeveen wrote: >>> >>>>On September 25, 2003 at 09:15:09, José de Jesús García Ruvalcaba wrote: >>>> >>>>>On September 25, 2003 at 08:26:21, Vincent Diepeveen wrote: >>>>> >>>>>>ICGA asked me to Call for participation in the world championship 2003. >>>>>> >>>>>>So far only 3 programs subscribed to join the world championship computerchess. >>>>>>DIEP is one of them, i guess Brutus the other one. >>>>>> >>>>> >>>>>And there three types of people: those who can count and those who can not. >>>>>José (: >>>>> >>>>>P.S. Good luck and lots of fun the world championship! >>>> >>>>Ah just had email from Stefan, they all didn't know you had to register 'so >>>>soon'. >>>> >>>>Usually world champs most things get organized at the tournament day itself, >>>>i remember especially the paniccing phase 1 short before the world champs >>>>started in October 1997, Paris :) >>>> >>>>Not a single organizer there from the home organisation (so not ICGA) spoke a >>>>word English (and my english isn't that good either, for sure in 1997 it was >>>>horror & co too), Dutch or German and my French is horrible, so i just sat >>>>down at a chair, installed my computer and just guessed what the hand movements >>>>of the home organisation meant. >>>> >>>>They must have guessed in advance to only receive French speaking participants, >>>>a normal assumption for French organisers :) >>> >>> >>> >>>That's really unfair. The 1997 World Championship organization in Paris was >>>great. Remember that we were playing not in some obscure university hall, we >>>were playing in the "Palais de la Bourse". >> >>There were good things and bad things. i bet it was great for french speaking. >> >>bad was all the neonazi demonstrations and the real long waiting times to just >>get past the 2 security checks each time. >> >>bad was not having any internet there or any other contacts to the outside >>world, i would not be able to imagine in 2003 to be without internet. >> >>publicity was real bad of the tournament. >> >>nothing online. really nothing. i remember thorsten czub phoning during the >>rounds to the outside world at his mobile phone the results, otherwise they >>wouldn't even know the results. >> >>So publicity was non existing. >> >>Considering the huge staff of frenchmen running in panic mode around during the >>whole tournament that was really a bloody shame, but what we would call here >>'typical french chaosmanagement' :) >> >>Bad was that it took so long to just get outside of the building to just get 1 >>small bread for example. >> >>Good was that each morning when walking to the tournament hall i could order for >>if i remember well 6 franc or something a big fresh bread at a breadshop. That >>tasted real good! >> >>bad was fact that there was still too many world titles then. There were just 3 >>competitors if i remember well for the professional world title. Virtual chess, >>CSTal and Fritz. >> >>Good was that after a few days there was very cheap drinks IN the tournament >>hall supplied by organisation. >> >>Good was that it was possible to walk around without problems in the tournament >>hall, in 2001 maastricht for example i found the location a bloody shame. They >>corrected that great in 2002 though in Maastricht. > > > >OK, so in short the good parts were the french bread and free drinks and that it >was possible to walk during the rounds. > >If you like the french bread you should try our chess programs. Some of them >would give you a good run for the money. ;) Well meeting each culture at a world champs is real important and leaves an impression that lasts forever. But the french bread,.... everyone must try it :) But the Tiger, well perhaps it plays me to french bread, so let's not try it too soon :) > > > > >>>The main problem I remember there was AMD. They sponsored the event and provided >> >>Oh the hardware, well Kallisto was sponsored a PII300 by intel. Kallisto however >>was still 16 bits and way faster (like 50% or some insane big diff) on the >>233Mhz K7 that Jan Louwman had managed to get too. >> >>So officially Kallisto ran on that PII300 but in reality diep ran at it. You >>couldn't get that cpu in any shop at that time. >> >>When i entered with that machine the tournament hall i was amazed to see that >>half the tournament was carrying a PII300 with him :) >> >>The toledo2000 programmer was not understanding that his DOS just went up to >>64MB hashtables instead of the full 128MB that he had on the PII300 machine :) >> >>But i guess the real bad thing from world champs 1997 in my memory is fact that >>only search depth mattered there really, assuming a debugged program. I lost >>game after game thanks to simple tactical errors. 8 ply search for a few moves >>and DANG opponent starts smiling "i win a piece!". > > > >Ah well, so that can be put in the "good" column I guess: you discovered there >that in chess your are helpless if your opponent outsearches you significantly, >and that you'd better debug your program before the start of the tournament. it was tested in 1997 with 1000 games auto232 by Jan Louwman. Without that i bet it would have crashed like so many others :) Well crashing is part of the fun... Nowadays some find it bad to crash, but it's part of life. Ask the winners of the previous 2 world champs. I remember they crashed last 18 games they played there around 30 times? >You could have discovered that at home, but well. Wisdom is wisdom. You don't want to miss the fun 500 processors are going to give in that respect. In fact i might get 1 or 2 testruns at 500 processors for 1 hour or so. Coming monday morning very early i will have 1 such testruns (those get scheduled automatically but because the machine is never empty some administrator must clean by hand the whole machine and let this run, so practically it only happens when they have maintenance at the machine) so that's very interesting. Some countrymen of yours can't in fact wait for that output. Expected efficiencyspeedup is a bit less than 37.3%, so you can do the match with 125GB hashtables at 1 position from Nataf - Svidler (Fressinet) As you know i managed to help Nataf in FIDE world champs by 'psychologically motivating' him. Hopefully this 500 processor run helps too: http://www.nao-cc.com/naocc/index.html That is what it will analyze some line that is in the 'psychological' interest of my friend :) > > > > >>In RGCC at the time some people like Bruce and Bob just posted their believe >>that search depth mattered and the rest was not important at all, especially a >>good evaluation wouldn't matter at all. >> >>Well how they were proven wrong later of course. > > > >As far as I know nothing has been proved, either way, in this area. That is not real true. Please delete your current evaluation function and replace it with tiger 1.0 evaluation functions, preprocessing etc. Just make the search the same like it is now and go compare. I bet that it will not kick any commercial engine that tiger 1.0 when compared to the current version. >The strength of computers at chess comes from a combination of good search >techniques and reasonable evaluation, with some emphasis still today on search. >Don't forget what you have learned in 1997... No that's not true. You *can* get another 5 ply easily by using a dumber evaluation function type Cilkchess. To quote Peter Gillgasch: "everything gives a cutoff in that case". What is more important, that 5 ply extra or some big penalties for king safety (to just give exampe of what cilkchess didn't have). It got like 17 ply without problems world champs 1999, but it was not impressive at all. Some programs it outsearched 8 plies. > > > > >>I remember Ernst running around the tournament hall crying loud that he hit 1 >>million nodes a second. >> >>A statement done by Peter Gillgasch over the email (he has programmed the >>darkthought in alpha assembly, i guess he knew more from it than Heinz still >>does) was very correct, and still stands in 2003: "darkthought can search so >>deep thanks to a very simple reason; it is positional so stupid that nearly >>*everything* gives a beta cutoff to it". > > > >So stupid that a bunch of programs were helpless against it. >What is your definition of stupidity? Achieving effectiveness with means that >are not to your taste? Right, as a real chess player i definitely appreciate that definition. > > > > >>In 1999 i joined with a diep version which was very dumb in endgame and i was >>really amazed to see search depths of 20 ply in endgame there when running at >>bob's quad xeon 400Mhz with a 400MB hashtable (which was real big to me in >>1999). Only end 2000 and especially 2001 and right now i'm improving endgame >>quite a lot, still ran behind there :) >> >>>computers, but there were not enough for everybody and some of them were >>>apparently defective (The K6-233MHz was new at the time and I remember Amir Ban >>>had been provided with a defective one and had to switch machines after >>>detecting a problem). >> >>>BTW I won't go to Graz. >> >>I can understand. Will you join dutch champs though? > > > >YES. I'll not be there in person but Jeroen is going to come. >We have a title to defend. Cool! > > >>> I guess the organizers wanted to prevent any computer >>>overheating problem by organizing the tournament in November in Austria? :) >> >>Well a tournament near the end of the year has something magic always. When it's >>in the middle of the summer you always miss the summer completely. Working in >>your room last minute, or being put in a room that's hot because the machines >>heat up the room bigtime and outside temperature also is usually very hot. >> >>Thanks to that you miss a part of the summer completely and the sun already >>shines here so little :) > > > >Sun all year long here... Perhaps i'm more motivated to work hard in order to see the sun :) > > > Christophe
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.