Author: Robert Hyatt
Date: 18:33:47 06/13/01
Go up one level in this thread
On June 13, 2001 at 18:12:58, john rice wrote: >Hello, >Could you please answer these 2 questions? >1. I was going to add all of the missing 5 piece endgames to the folder named >"tablebases" my Nalimov endgame turbo program created. So I went to Dr. Hyatt's >website for the files. The problem I discovered is the endgame files have a >slightly different name. For example: the Nalimov file appears like this >(kpknbw.emd) while the files on Dr. Hyatt's site appear like this (kpk.nbw.emd). >Will these files work together? Why the . between kpk and nbw? So far as I know, there are two acceptable names: kpk.nbw.emd, or kpk_nbw.emd. I have no idea what has happened with the particular program you are using. For crafty (and most public programs that use Eugene's probe code) the above will work. > >2. I am running a dual harddrive (raid) setup. It is twice as fast as a single >harddrive. I run raid here on several machines. I don't find it to be any faster at all, for normal cases (raid is definitely slower in writing). The speed of the disk is not as much a problem as the speed of the PCI bus the data has to move over. I don't see how you can get 2x the performance with any raid (raid0 or raid5). > I would like to take advantage of the access speed these harddrives >provide by increasing the depth of assessment of the tablebase. In the engine >settings for Deep Fritz the tablebase depth setting has values of 0 up to 20. >The default is 3. What setting would you recommend for optimal strength for the >endgames? I believe the lower the number the deeper the assessment. >Thanks in advance, >Jr I wouldn't venture a guess. Crafty is set up to run best with fast disks. IE I use 10K rpm 80mbytes/sec SCSI drives for my TBs. That is _extremely_ fast. I don't know how Fritz probes so I can't help at all there. Crafty adjusts itself automatically.
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.