Author: blass uri
Date: 15:29:18 01/19/99
Go up one level in this thread
On January 19, 1999 at 16:10:31, Robert Hyatt wrote: >On January 19, 1999 at 00:54:03, blass uri wrote: > >> >>On January 18, 1999 at 18:56:22, Robert Hyatt wrote: >> >>>On January 18, 1999 at 07:15:07, Jouni Uski wrote: >>> >>>>Sorry, but I haven't it here in our office. I take it tomorrow with me, if >>>>nobody posts it before. >>> >>> >>>Just a comment as I have said this before. At present, based on a _lot_ of >>>games vs other programs on ICC, I believe that Junior 5 is probably the most >>>dangerous program around. At least when playing computers. I haven't watched >>>it against humans very much. But against computers, including mine, it is >>>_very_ strong... >> >>I do not follow ICC. >> >>1)What are the results which you are based on? >>2)Do you play against Junior5 or against Amir Ban's latest version? >> >>Uri > >Both. "ban" is (I assume) the latest version they want to test. Lonnie (and >others) use junior 5 all the time. I am not doing badly against Junior at all, >but when you factor in that my machine is around 4x faster than the hardware >either is using. Which gives me a _really_ serious speed advantage (IE I am >averaging about 700K-800K nodes per second on this machine most of the time.) > >Of all the programs I play (Lonnie seems to have 'em all) Junior 5 and 'ban' >seem to be the strongest by a significant margin. IE against ban, just for >the month of January (about 1/2 over) crafty has won 22, lost 11, drawn 18. > >I'd hope that my effective speedup of 2x-3x would produce a larger win/lose >ratio than that. I think that speedup of 2x-3x produce similiar win/lose ratio. If I look at the ssdf rating then I see difference of 60-100 elo for the same program when the hardware is 2-3 times faster. Your result give you performance of near 400*(22-11)/(22+11+18)=4400/51 elo points more than Junior5 so the result is readonable if I assume equal programs. But it doesn't. Now if you believe, as I do, that faster >hardware always improves a program, then the above is explainable and I would >conclude that maybe Junior is stronger, _if_ it has equal hardware. If you >believe (as some do) that we have reached 'tactical sufficiency' and faster >hardware isn't important, then this result means something else entirely. I'm >in the former camp, which I believe to be correct here. It is clear that hardware is important because it is easy to see that all the programs in the ssdf list(in cases they play on pentium200MMX and on P90) have better results with pentium200MMX I do not know about people who believe that faster hardware is not important 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.