Author: Sune Fischer
Date: 00:44:45 04/16/02
Go up one level in this thread
On April 16, 2002 at 00:35:06, David Dory wrote: >On April 15, 2002 at 06:08:21, Sune Fischer wrote: > >------------------------------------------------------------------------------> >>In theory bigger is better, in practise you don't get much improvement beyond a >>certain size. > >You're thinking of hash sizes. If you can't use a LOT more ram memory, HUMONGOUS >hard disks, BIGGER cache(s), and FASTER EVERYTHING, in your program, your >programmer is brain dead, pull the respirator! >------------------------------------------------------------------------------ So the Tiger on a Plam OS must be brain dead? I don't mean to insult you, but you have never written an engine, have you? >>do you have any idea how Christophe got Tiger 200 elo stronger than Crafty? They are both using the same alpha-beta, so how >>could one possibly be 200 elo stronger, explain that! >------------------------------------------------------------------------------ >Simple: > >Christophe has put in a great amount of time and effort and testing and has >found ways to better tune his evalu8, his extensions, futility pruning, >further optimized his code for speed, added more chess knowledge, etc. > >And all that work won't equal the improvement Chess Tiger will get when he's >able to re-compile and optimize his program for the next (faster), CPU and >system, later this year. I don't believe they are at the hardware limit just yet, it may come at one point. AFAIK Chris said Tiger had improved greatly since he started working with Ed (was it?). I can't find the interview just now... >You won't SEE that gain compared to other engines because (guess what), ALL the >top engines will be stepping up to the latest hardware, also! You state the obvious. >Thus the HARDWARE IMPROVEMENT for chess programs may not be so well noticed, but >I assure you - it's there, and it is the major factor in the improvement of >chess programs in the last 20 years. >------------------------------------------------------------------------------- Hmm, so explain this: AMD K6-2 400 MHz: Fritz 5.32 - 2552 Fritz 6 - 2623 ------------------ elo progress in development: 71 points Deep Fritz: K6-2 450MHz 128 MB hash - 2654 Athlon 1200MHz 256 MB hash - 2726 ----------------- elo progress by 3.5 (and NOT 2) times better hardware: 72 points How many years was it between the K6-2 and the Athlon 1200, and how many between Fritz 5.32 and Frtiz 6? I'm pretty sure software is a pretty big factor here, DAVID! >>The countless number of hours put into testing and analysing, you don't think >>that is going to pay off? > >The software will improve - but slowly, much more slowly than the hardware. I >don't care how many "hours of testing and analysing" the programmer's do. talk talk talk, show us your data please. >You don't think hardware engineers aren't spending "hours of testing and >analyzing"? > doh? >>Hardware _is_ a factor, but not the only factor, which you seem to think. > >THAT'S NOT WHAT I WROTE: I said hardware improvements were the cause of MOST of >the improvements in chess program strength in the last 20 years. > >I never said "All the improvements were due to hardware." > >I'm disappointed that you wrote the above when clearly I didn't say that. You didn't say anything clearly, most of it was rubbish. >If you believe that "top programmers break new ground every day", well I've got >a "shovel" and will be glad to re-introduce you to the "ground". It's very hard, >and doesn't move when you yell at it "get out of the way". <grin> > >All this "break new ground every day", stuff is one-half hype, which has to be >put into context. no hype, the tree is huge, plenty of places to cut, just have to know where. >A programmer may "break new ground", but that doesn't mean it will improve the >strength of the program, even a little bit. The "new ground" will be tested, and >many times goes right into the shitter! And many times it won't.... >It takes a lot of work just to get an extra 25-50 elo out of a good program. > >Meanwhile, HARDWARE is improving by leaps and bound. Less than a factor of 2 a year, which is less than 50 elo, which is about what they can do in software (judging by the Fritz progress). >>So I take it you still believe programs are 15,000 times stronger because of >>faster harddrives? >>Well, at least I tried... ;) >>-S. > >NEVER SAID THAT, Oh, so you meant NOTHING but NONSENSE with those numbers, why post them then, DAVID? (oh no, I've been infected with the troll caps-lock disease, must go see doctor soon...) >I do wish you'd follow the discussion more closely. I stated quite clearly that >no program could use these system hardware improvements, fully, but that >HARDWARE is responsible for more chess strength improvement in the last 20 >years, than software. I have to disagree when you put it like that. I think it is closer to 50-50. >David -S.
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.