Author: José Carlos
Date: 14:34:50 06/04/01
Go up one level in this thread
On June 04, 2001 at 14:52:51, Uri Blass wrote: >On June 04, 2001 at 14:24:53, José Carlos wrote: > >>On June 04, 2001 at 14:03:45, Rajen Gupta wrote: >> >>>On June 04, 2001 at 12:41:14, José Carlos wrote: >>> >>>>On June 04, 2001 at 09:35:41, Rajen Gupta wrote: >>>> >>>>>On June 04, 2001 at 08:57:09, José Carlos wrote: >>>>> >>>>>>On June 04, 2001 at 08:28:43, Jouni Uski wrote: >>>>>> >>>>>>>1/2/(5) years? Yes, if we remove benefit from better hardware, what is left?? >>>>>>>Any real improvement is playing level of top engines. 5 points from Nalimov >>>>>>>may be?! >>>>>>> >>>>>>>Jouni >>>>>> >>>>>> I think you cannot remove benefit from better hardware, because most >>>>>>programmer adapt to new hardware to get better programs. For example, a lot of >>>>>>programs have become SMP. This has happend because multiprocessor machines have >>>>>>become cheaper, so it is a hardware improvement that leads to software >>>>>>improvement. >>>>>> Besides that, you can chose your favourite program and compare versions. Let's >>>>>>say Fritz. Compare strength of versions 4.01, 5.32, 6, Deep Fritz. I think you >>>>>>can answer your own question, can't you? >>>>>> >>>>>> José C. >>>>> >>>>>i'm not sure: see my previous post. get the chessbase engines to play each other >>>>>from an individual copy of the same opening book-you'll be surprised at the lack >>>>>of any difference in engine strength.i'm planning to play the 2 tigers vs j5, >>>>>f5.32 and f6 light, all from individual copies of the same opening book(tiger >>>>>book).also going to play f5.16 vs f4 vs h6)if there is no significant difference >>>>>in playing strength between them (i have a suspicion that it might be the case) >>>>>then i'll probably stop buying newer programmes. >>>>> >>>>>inmo 95% of the so-called increse in strength is purely the result of better >>>>>opening books, book learning functions, access to endgametables and tuning >>>>>against the immediate predecessor programmes. >>>> >>>> I'm sorry but I'm not gonna argue about that, since IMO the opening books, >>>>learning functions, endgame tables etc... _are part of the engine_. So the >>>>discussion does not make sense for me. >>>> Just one suggestion: why don't you also disabe hash tables, null move, >>>>pondering and alfa-beta to test the 'engine'? These are parts of the engine, >>>>_exactly the same_ as opening books, learning functions, etc... >>>> I'm really very surprised about this discussion about the opening book not >>>>being part of the engine. >>>> I suggest to let the programmers define _what is the engine_ and what are >>>>peripherals. Or do you discuss with your car's mechanic _what is part of the >>>>engine of the car_ and what is not? >>>> >>>>>i doubt one would get a better analysis of a data base from the newer programmes >>>>> >>>>>rajen >>>> >>>> Just try it. Don't doubt, test. >>>> >>>> José C. >>> >>>i dont quite know what you mean by all these things. the engine is purely the >>>software that can calculate the chessmoves and the opening books is an opening >>>database that a chessengine refers to when playing the first few moves of the >>>game. any chessengine can use any database and i'm playing one engine vs another >>>using the same opening database(which incidentally is separate from the engine. >>> >>>this will purely test the engine improvement and i doub that there is anything >>>significant to tlak about >>> >>>rajen >> >> Ok, if you don't want to understand, it's your choice. >> >> josé C. > >I am going to explain the point. > >People are using chess programs to analyze chess positions the question is if >the new engines are better in analyzing positions then the old engines. > >Opening book may be relevant to the playing strength of the program but they are >not relevant for analyzing a middle game position. > >Tablebases are relevant for the playing strength of the program but they are >irrelevant for analyzing middle game positions when there are no tablebase hits. > >things like hash tables,null move are clearly relavant for analyzing every >position so I see no point in disabling them. > >Uri I know you understand it, Uri. I do understand it too. But people seem to not want to understand. This is not for you, but for people who refuse to understand: - One thing is a chess program: a chess program is made up by a GUI and an engine. The engine of a chess program has many parts: move generator, opening book, search algos, hash tables, endgame tables, pruning heuristics, etc... - Another thing is an anylizing module. It is a small part of a chess program. For example, pondering, book opening, time management... are not part of an analysis module. An analysis module does not need to be able to play a game of chess, because it's not intended for it. Most of us, chess programmers, intend to write _chess programs_, not _analysis modules_. So opening books are part of our engines. If you want to compare analysis modules, it's wrong to take Tiger or Shredder, because they are written as _chess programs_. The programmer of a chess program writes his book and his evaluation knowing they'll work together. Chris (or Jeroen, or whoever) puts aggresive moves in Tiger's book, because Tiger's eval is aggressive, and so on. So please (I'm so tired of repeting this...) don't separate important parts of chess programs. When a programmer releases an analysis module, and he states it, then you can compare it to other analysis modules, without opening books, time management, etc. I hope this time I'm clear enough. José C.
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.