Author: Robert Hyatt
Date: 10:27:05 07/21/00
Go up one level in this thread
On July 21, 2000 at 11:27:52, pavel wrote: >On July 21, 2000 at 11:04:03, blass uri wrote: > >>On July 21, 2000 at 10:50:00, pavel wrote: >> >>>mainly to Bob, >>>but anyones else's help is appreciated. >>>I am worried about crafty's speed. I think in a single processor, the only thing >>>that pulls it behind the commercial programs, is the speed. In a certain >>>position in my hardware when a commercial program has 13 depths or atleast sees >>>deeper in less time, on the other hand crafty takes "comparably" longer time to >>>see the same position or better position. I am comparing between the chessbase >>>programs (junior, fritz) and crafty (as a chessbase native program). >> >> >>The depth is not important because different programs do different extensions. >>If you compare Crafty's depth with hiarcs you will find that Crafty goes deeper >>but Hiarcs also wins Crafty in blitz. > >If you go to chess.net you will see some some fritz accounts (example >testing..., fritz-x) they use and autoplayer to play fritz on online chess >servers. watch fritz clones play against vipp-computers and also watch it play >against bryant. you will know the differance. Bryant runs on a single processor >and so doesnt see the same "depth" or "plies" as fritz does. but when it plays >against vipp-computers it either plays the same level or worse. but not better. >i am not coming to any conclusiong with this example. >but saying that with same speed ( or same depth) crafty plays better comperably, >as vipp-computers runs on multi-processors. > > > >>Junior's depth is not plies so you cannot compare apples with oranges. > > I am not comparing anyone with anyone ..... >I am talking about crafty's playing strength with increased speed. >the chessbase programs was an example..... >and I stated that too. > > >> >>I do not see a reason to worry about the fact that commercial programs win >>Crafty in blitz. > >I do not worry about commercial programs winning crafty in blitz. >what I said was crafty looses commercial programs because of its speed most of >the time. or atleast thats what I meant. > Crafty could be significantly faster. The reason it isn't is that making it faster makes the code far more complex and more difficult to change. As things are "solidified" I do tend to optimize them. IE the rotated bitmap code is pretty lean and mean. But the evaluation is continually being modified, and in software engineering "Design for change" is more important that "design for speed" when "change" is a common thing to do. It is likely that it could run maybe twice as fast with a _lot_ of work, and I am not talking about assembly language at all. Cray Blitz was _highly_ tuned to the cray architecture. It was also highly difficult to modify, unless you only wanted to tweak a weight here or there. I don't want to go down that road with crafty as there are too many things left to try, and I want to make it as easy on myself as possible to try them... That's why it is slower than it ought to be, but it is not a 'slow-poke' by any stretch... > > > > >> >>Crafty wins faile in all time controls(You can download fail in the chessbase >>site) and I do not see it as a problem. > >apple and oranges?? you are comparing crafty with faile?? >> >> >> >>>not many people has a multi processor. So I am wondering if is it possible to >>>maintain the same quality of game but with better speed? which is also my main >>>question. >> >>Bob does not do it and the reason is simply that he does not know about a way to >>do it. > >dunno wbout it but thats what I wanna know? ;) > > >> >>I am sure that Bob is interested in doing his program better. >> >>>many people in this forum come up with blitz matches (5 min mainly) between >>>crafty and chessbase products, and claims that "this" program is far better than >>>crafty. >> >>The ssdf does tournament time control games and Fritz6a is number 1 and not >>Crafty(not every program can be number 1 and I do not see the problem). >again I dont know whats your point. > >but as I have said i am not comparing anyone with anyoneelse here. >IMO SSDF is not perfect, also its "bogus" ( not literally) >but again its only "my opinion" > > >I thought of a better feedback. >seems like anti-crafty >anyways thats for the info. ( specially the ones about junior and apple and >oranges). >pavel > > >>Uri
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.