Author: James T. Walker
Date: 17:39:24 12/17/98
Go up one level in this thread
On December 17, 1998 at 16:09:11, Dr. Gregor Overney wrote: >Interesting game between two computer chess programs. Since you are using two >different computers (PII-333 and K6-2-350) I am wondering if you make >adjustments to consider the difference in computer performance (e.g. adjust the >total time for 40 moves). Are you using PC-100 memory for the K6-2-350? What >about the disk access? Are you using large books that require frequent disk >accesses? Any table bases? Are you using the same disks and disk controllers. > >I am sure that you have already mentioned those details but I cannot find the >post. I also remember that you talked once about time adjustments when comparing >two chess programs on different computer architectures. But I do not remember >the equation you suggested to make this correction. > >Gregor >>>>>>>>>>>>>>>>>>>>> Hello Dr. Overney, I think you have me confused with someone else. In any case I did not make any adjustments. As a matter of fact I'm very confused about what would be a proper adjustment. The "Marks" test for Junior and Nimzo and Fritz do not make much sense to me. For instance on one computer the KNPS is higher but the Nimzo mark/Junior mark/Fritz mark is lower with both using the same hash size. I don't think the speed difference is enough to worry about anyway. The difference between the two computers is at most 25%(Usually around 10%-15%) depending on what test I try. The K6-2 has 100 Mhz ram and the PII 333 has 66 Mhz ram. I think the difference for even 25% speed difference would be only 10-15 rating points. The books were the ones that came with the program. I think Nimzo 99 book is bigger than Junior. Both books are on my hard disk and the access time is "instant". I'm not letting the programs use the Fritz tablebases because I only have one CD and it would not be fair to the other program. The Nimzo 99 program has an "Opponent" function that is supposed to make the program play better against certain programs (Fritz,Junior & Rebel) but the auto232 function will not allow that part of the program to function. The subject game was played with the auto232. I have run about 60 games between the two programs (With various time controls) and Junior has a very slight lead. All games were run with the auto232. The Nimzo 99 upgrade I downloaded does seem to make the strength closer to Junior. Still not sure which is stronger but so far Junior has won all three games at 40/2hrs. Jim Walker
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.