Author: Chessfun
Date: 16:14:32 02/03/00
Go up one level in this thread
On February 03, 2000 at 19:00:58, Christophe Theron wrote: >On February 03, 2000 at 17:42:59, Chessfun wrote: > >>On February 03, 2000 at 14:43:13, James T. Walker wrote: >> >>>Yesterday I played Rebel Tiger vs Fritz 5.32 some Bullet/blitz games to try to >>>get a feel for which is stronger. I admit my format is a little strange but the >>>results were interesting to me anyway. I only have a PII-333 and a K6-3-450 so >>>I have to swap the programs for half of the games to be fair. This time I tried >>>something a little different. I gave the program on the PII-333 3 min/game >>>while I gave the program on the k6-450 only 2 min/game. The first 110 games >>>Tiger had the K6-3-450 and Fritz 5.32 had the PII-333. The score of the first >>>110 games was 60-50 favor Tiger. Of course I changed computers and tried again >>>for another 110 games. The second score was 59.5-50.5 favor Fritz 5.32. So the >>>totals were Chess-Tiger 110.5 to Fritz 5.32 109.5. I doubt if anyone wants to >>>see the 220 games. One more thing, I played Fritz 5.32 without access to it's >>>endgame CD. I thought it interesting that 50% more time failed to compensate >>>for the 35% speed advantage. Maybe not enough games? >>>Anyone care to comment? >>>Jim Walker >> >>Hi Jim. >> You should try this with H732. I once let it and F532 play >>1600, one minute games. H732 scored over 1300 !!. >>I wasn't proud to admit I run these types of games so I never >>posted the result. It was just after someone here made a statement >>on playing 1000 one minute games between these two programs. >> >>Let H732 try the Tiger....I would be interested in those results. >> >>Thanks. > > >Yes, it's interesting. > >However, as time controls get faster and faster you must take special care about >GUI issues. > >The Rebel-Tiger GUI, like any GUI, eats a little bit of time on every move. The >time is used to update the graphical board, update the toolbars, the clocks, >send an optional sound, and so on. > >If it is possible you should disable as many features of the GUI as possible. In >Rebel-Tiger, that means: >* turning sounds OFF >* closing all the windows (leave only the chessboard) >* close all the toolbars >* maximize the program >* set the engine to high priority >* don't give too much hash tables. In 3mn/game, 4Mb is enough. In 1mn/game, 1Mb >is enough. > >And: don't move the mouse during the games (every time the mouse moves, a lot of >messages are sent by the Windows system, this takes a lot of time). > >In your experiment Hiarcs against Fritz, are you sure the GUI did not handicap >Fritz? Didn't you give too much hash tables? In very fast games, just 1 tenth of >second lost per move is a killer. > > > Christophe It is possible hash was set at 4 mb, I ran the match in both Fritz 5.32 GUI and Hiarcs 7.32 GUI. As for some reason every 250 games or so it stopped. I only noticed these games today when looking at matches and tourneys to post at my website. I noticed that ponder was off in both GUI's as the title of the matches is 1min4poff, this was at a time when pondering on or off in those interfaces was being talked about, now with F6 it is much more clear. I ran this months ago but don't remember anything odd with it, I will check later after current 40'/40 40'/40 40' tourney is finished. Funniest thing I noticed as I went through my databases was some 30 min matches with I think F5.32 and H7.32 where Hiarcs won at least 8 games in the same variation. Whenever it had white they both went at the same opening again and again. Eventually F5.32 after 8 losses got a draw then they changed openings again. I ran these two engines at almost every time control possible over the last year. As I said in the last post I never mention the fast fast ones lol. Thanks.
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.