Author: Joshua Haglund
Date: 11:33:28 01/24/03
Go up one level in this thread
On January 24, 2003 at 10:09:31, Andrew R. Case wrote: >On January 24, 2003 at 08:59:28, James T. Walker wrote: > >>On January 24, 2003 at 01:04:17, Brian Katz wrote: >> >>>In the following famous position, which I am sure many of you are familiar with, >>>Botvinnik - Capablanca, AVRO Holland 1938 1-0...White (Botvinnik) played the >>>extraordinary 30.Ba3!! and went onto win Brilliantly. >>>The remaining moves were:30...Qxa3 31.Nh5+! gxh5 32.Qg5+ Kf8 33.Qxf6+ Kg8 34.e7 >>>Qc1+ 35.Kf2 Qc2+ 36.Kg3 Qd3+ 37.Kh4 Qe4+ 38.Kxh5 Qe2+ 39.Kh4 Qe4+ 40.g4 Qe1+ >>>41.Kh5 Black Resigns 1-0 >>> >>>[D]8/p3q1kp/1p2Pnp1/3pQ3/2pP4/1nP3N1/1B4PP/6K1 w >>> >>>My question is will running full games with Fritz 7......................... >>>Hash Tables set at 818 MB cause heat problems for an AMD Athlon 2600+ 1 gig of >>>DDR SDRAM. >>> > > >>>I have beeb experimenting with Fritz 7 Hash Table settings in the infinite >>>analysis mode. >>>Here are some results based on various Hash Table settings. Time to find 30.Ba3 >>>as a winning move not just to see it in the list of alternatives. Also nodes per >>>second. >>>At 818 MB Hash, I hear no extra work on the harddrive, only when first setting >>>the Hash at 818 for a few seconds then no more. >>> >>> >>>The results from 16,32,64,128,256,and 512 MB Hash Tables respectively were all >>>basically the same. >>> >>>At 16MB it took 3 seconds to notice that 30.Ba3...=(-0.16) and Black is better >>>and then it drops to 0.00...Then at 1:33 it jumps to + 0.34 in favor of White. >>>Then at 3 min 15 secs.it jumps to +-(3.97) kN/s at 1116. >>> >>>At 256 MB it took 3 seconds to notice that 30.Ba3...=(-0.16) and Black is better >>>and again drops to =(0.00)...Then at 1:05 it jumps to + 0.34 in favor of White. >>>Then at 2:00 it jumps to +-(3.31) at 1064 kn/s >>> >>>Now at 512 MB everything is basically the same except the value at 1:02 is at >>>+0.50 rather than the + 0.34 as above. Then at 1:57 it goes to +- 3.88 >>> >>>At 768 the only real difference is that the value at the 1:00 minute mark is >>>down at + 0.34 >>> >>>At 818 I get similar results early on then at 1:03 + 0.50 which is higher than >>>the reading at 768 MB setting, and 2:08 the Highest value at this early stage +- >>>(3.97) kN's flucuate in the low 1000's in all of the settings listed except when >>>set at the 16 MB Hash which was in the 1100's. >>> >>>So basically, will the higher value in the same time period such as the + 0.50 >>>as opposed to the + 0.34 make a difference in a game. Will the higher Hash Table >>>818 MB settin, yielding an evaluation + 3.99 at 2:10 be better than the + 3.88 >>>at 1:58 with the lower Hash Table setting? >>> >>>My main concern is that I do not want to over tax my processor. >>> >>>If anyone has some insight on this It would be greatly appreciated. >>> >>> >>>Thanks >>>Brian Katz >> >>Hello Brian, >>How do you get Fritz 7 to accept 818M hash setting. Anything I set over 288M >>seems to default to 287 as noted in the "Notation" window where the moves are >>displayed. >>Jim > > > The reason you can't get hash tables larger than 287mb is your >operating system. Only windows nt, 2000, or xp can make use of larger hash >tables. I find it's from the amount of ram... a certain percentage, say 75%
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.