Author: Slater Wold
Date: 11:36:10 01/16/02
Go up one level in this thread
On January 16, 2002 at 12:33:09, Thomas Mayer wrote: >Hi, > >>I don't have corrupt TB's, but of course the 6 man is not complete. They are >>not all available yet. > >afaik Junior 7 does not use 6 men at all in search... Yes it does. I setup a position with 6 pieces (a TB I had) and it was 100% TB's. >but strange is, that you have different results in score, nodes etc... so this >example does not prove anything... Sure it does. The faster search makes the score update quicker. And the scores are exactly the same. >it seems that the hashtables were not deleted Yes they were. That is an option on DJ7. >or whatever, in common you should get same scores, nodecounts, egtbhits etc. - I was using two different HD's. Like using 2 different CPU's. Should they match up then?! >ah, and you shouldn't try to comparisson with Deep Junior to measure the disk >speed, with more then one CPU the results are not to reproduce... Yes they are. I tried that after my first test. Using 2 CPU's is sproadiac, but not on the same position. Although SMP branching is mostly random. >maybe you try >it with a winboard engine that definitely clears hash after every setboard The hash was cleared. I had to close the program completly to do this test, and I had DJ7 clear the hash. It's an option on the engine. >command. One I know that for sure is Quark... :) The current release version >uses TBs VERY agressive so that might be a good disk test... (Please use my >version then and not Danns compilation - Danns is faster but does also produce >unreproduceable results for whatever reason...) Ok, I will when I get home. > >Greets, Thomas
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.