Author: James T. Walker
Date: 20:31:47 08/15/99
Go up one level in this thread
On August 15, 1999 at 13:43:29, Mark Young wrote: >On August 15, 1999 at 12:15:34, James T. Walker wrote: > >>On August 15, 1999 at 04:38:32, Terry Ripple wrote: >> >>>On August 15, 1999 at 03:43:41, Keith Kitson wrote: >>> >>>>From the discussions here I get the impression that there must be some research >>>>that can be done to generate a table showing reasonably optimium hash memory >>>>sizes for a given time allocation and speed of processor. Has anyone done any >>>>research in this area? >>>> >>>>I suspect for anyone that is not aware of the hash/processor speed/time control >>>>ratio affect, games played with Hiarcs732 may not always be running Hiarcs at >>>>its optimum settings. >>>> >>>>I would expect in future relases of strong programs the author should give more >>>>guidance on optimum settings for the parameters mentioned above, where the >>>>program strength could be affected. Better still how about an intelligent >>>>algorithm that is labelled say 'strongest setting' that determines the speed of >>>>processor, time controls set and either automaticallky sets or suggests the >>>>optimum hash for the determined parms? >>>> >>>>Do we have any programs that do this at present? >>> >>>I believe Fritz does this accuratly but Hiarcs7.32 don't work properly in the >>>Fritz interface. >>> >>>Regards,Terry >>>> >>>>Keith >>Hello Keith, >>I believe you are wrong about Hiarcs. Forget the HiarcsMark. You can only use >>it to compare your results vs other machines when using the same size hash >>settings. Mark (Hiarcs Programmer) said when playing a game with Hiarcs, please >>give it all the hash memory you can. I have found this is true. I'm currently >>playing Hiarcs with 200 Meg of hash table memory at Blitz. It works great! I >>have currently played 988 blitz games of Hiarcs 7.32 vs Fritz 5.32 on equal >>hardware (2 computers). The score indicates that Hiarcs has a 12 elo point >>advantage over Fritz. So they are too close to call and there is nothing wrong >>with Hiarcs. > >I am also running Hiarcs 7.32 with max hash at all time controls. It works fine >and does not hurt Hiarcs 7.32 at all and even helps it some I now believe. Your >results pretty much mirror mine, I also show Hiarcs 7.32 better then Fritz by >about 18 points in head to head play, but with fewer games only about 200. When >I calculate all game played by Hiarcs at all time controls against all other >programs. Hiarcs 7.32 elo advantage is almost 50 points better then Fritz, do >you show the same. Hello, No, my results from over 3600 games has Hiarcs only 3 points above Fritz. Fritz has played 1786 games and Hiarcs has played 1578 games so far. This is all at blitz times like 3/1 4/0 5/0. 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.