Author: Vincent Diepeveen
Date: 15:22:52 04/01/03
Go up one level in this thread
On April 01, 2003 at 17:41:30, Keith Evans wrote: >On April 01, 2003 at 17:02:55, Vincent Diepeveen wrote: > >>On April 01, 2003 at 15:24:17, Keith Evans wrote: >> >>>On April 01, 2003 at 14:48:03, Slater Wold wrote: >>> >>>>Correct me if I am wrong, but this is Dr. Donninger you're speaking of correct? >>>> >>>>I thought that was the name in the e-mail Bob posted, but I could be wrong. >>>> >>>>And, just so you know, you Bruce are a better man than me. I would have gone to >>>>Graz, and done some wall-to-wall counseling. >>> >>>All the more reason for you to come up with a Brutus killer ;-) >> >>I'll show up with europes fastest machine for computerchess for that >>purpose. From which i can use about 500 processors. >> >>So they gotta win that war quick, i fear the machine might be in use otherwise >>:) >> >>This time i have a year to prepare to play on that machine. Preparement is going >>very well! > >But Slater is from Texas... > >I will be interested to see how your beast does this year. thanks, my personal feeling is that remarks as made a year ago already by Arturo are very important. Namely Diep needs to get better tested and internet testing doesn't count much there. Note that internet testers in general leave you directly if they lose 10 games in a row. They kick too much on a high rating and go for some kind of freeware shredder or shredder itself to play online. At the moment i have 1 tester of diep kind of and he's having a break now too. I suck myself as a tester and such tests you cannot reveal of course as it shows more than just your openings preparation, which is already important to not show as when the opposing forces from austria deliver a book that kills you twice if your lines are inferior (book from Kure is not only in fritz but also nimzo and if it wins and it is in chessbase favour they give it to junior too for 1 game; that's how commercial world works of course). In short i feel that chessprograms have become that strong nowadays that you cannot afford major weak spots. Also not when you have 500 processor machine. that was shown just too clearly last so many tournaments. cct5 i had a hashtable bug in fact (incredible but true) which caused it to overwrite searches. In fact i found out about that bug only after i manually checked a game against Brutus and diep with a -2.0 score at 9 ply or so played into a helpmate in 3. That found the bug for me. Such bugs which were of course common to sometimes happen in the past (see the FHR reductions from Zugzwang or the beancounter only evaluation from cilkchess), but still could win you a world title if you had enough computing power. It is very obvious that nowadays you cannot afford to have a program that has a bug in doubled pawn code or has a bug in king safety code which causes it to castle into a mate (like diep - SOS in paderborn 2003, also a clear bug left into it because of bad testing). In that respect i will know already months in advance whether i make a big chance to win the world title. the thing that is for sure is that i cannot complain about search depth. this machine delivers 1 terabyte bandwidth a second. Just IMAGINE!! >Regards, >Keith
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.