Author: James T. Walker
Date: 06:21:32 09/22/99
Go up one level in this thread
On September 21, 1999 at 23:16:38, Joshua Lee wrote: > >>While the HiarcsMark test is affected by the amount of hash, it will give >>misleading results about what is best for playing. According to a post here >>some time back you should give Hiarcs all the memory available without swapping. >> I have run test of over 1500 games and I am sure that even at G/1 min Hiarcs >>does better with more memory. >>Jim Walker > >with 64MB of system ram Hiarcs says use between 40 and 48MB, this is all fine >and dandy but i have seen hiarcs quote en quote play better with 35MB, for >example it solves pos 14 on the LCT II test at around 4min compared to over 5min >with 47MB so so much for what Hiarcs says about optimum settings. I am more >concerned with deciding on the upcomming K6-3 500 on .18micron process or K7 >500, as i've said i have a K6-2 300 and i want to upgrade but i don't know what >is for 1 twice as fast and for 2 faster than the P3 550 (other than the over >priced K7 600 or 650. maybe someone reading this can shed some light on the >subject. I have been looking at the Benchmarks but i don't know if i should look >at FPU winmark or Cpu99 arrrgh. Thankyou Hello Joshua, Yes, you are correct. If you are doing the "HiarcsMark" test or testing positions then you will find that Hiarcs is faster with less memory than the max. I am talking about playing games. When playing, because of some unique way Hiarcs does not reset it's hash memory after each move, it plays stronger and is faster when it has maximum memory. I have proven this with many hundreds of games at G/1, G/5, G/15 & G/25 settings. 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.