Author: Antonio Dieguez
Date: 16:50:33 09/20/02
Go up one level in this thread
On September 20, 2002 at 18:20:56, Dann Corbit wrote: >On September 20, 2002 at 18:18:07, John Merlino wrote: > >>On September 20, 2002 at 18:00:48, Tony Werten wrote: >> >>>On September 20, 2002 at 16:45:28, John Merlino wrote: >>> >>>>On September 20, 2002 at 16:37:38, Will Singleton wrote: >>>> >>>>>On September 20, 2002 at 16:23:53, John Merlino wrote: >>>>> >>>>>>>1:52 13/17 -1.54 11346643 1...Ba4 2.Bd1 Kd3 3.e5 Kd2 4.Bf3 >>>>>>> Kc2 5.e6 Kxb2 6.Kg4 Kc1 7.Be4 b2 >>>>>>> 8.Kh4 Bc6 9.Bd3 b1=Q 10.Bxb1 Kxb1 >>>>>>>3:58 14/18 -1.56 23785267 1...Ba4 2.Bd1 Kd3 3.e5 Kd2 4.Bf3 >>>>>>> Kc2 5.e6 Kxb2 6.Kg3 Kc1 7.Be4 b2 >>>>>>> 8.Kh4 Bc6 9.Bd3 b1=Q 10.Bxb1 Kxb1 >>>>>>> 11.Kg4 >>>>>>> >>>>> >>>>>>Here's what my wonderful new P4-2.66 says: >>>>>> >>>>>>0:24 13/17 -1.56 9124305 1...Ba4 2.Bd1 Kd3 3.e5 Kd2 4.Bf3 >>>>>> Kc2 5.Be4+ Kxb2 6.e6 Kc1 7.Kg4 >>>>>> b2 8.Kh4 Bc6 9.Bd3 b1=Q 10.Bxb1 >>>>>> Kxb1 11.Kg4 >>>>>>0:42 14/18 -1.56 16053836 1...Ba4 2.Bd1 Kd3 3.e5 Kd2 4.Bf3 >>>>>> Kc2 5.e6 Kxb2 6.Kg3 Kc1 7.Be4 b2 >>>>>> 8.Kh4 Bc6 9.Bd3 b1=Q 10.Bxb1 Kxb1 >>>>>> 11.Kg4 >>>>> >>>>>Why are the node counts different? >>>> >>>>Because I'm stupid.... Well, not exactly -- let's just say I confused myself. >>>> >>>>We are working on a French version of Chessmaster, so I re-installed the program >>>>and the user that I logged in as was using the default of 4MB hash table size >>>>for the PIII-600 version, instead of the 32MB I used for the P4-2.66 version. >>>>Here is what the PIII-600 says with 32MB of hash (same node count, but >>>>noticeably faster than the 2MB version): >>>> >>> >>>I'm sorry, did you just say CM got faster from more hashtable ? >>> >>>I'm confused now. >>> >>>Tony >> >>Yes, I did. Why are you confused? > >Maybe he was thinking of NPS (which usually goes down) instead of time to >solution, which normally speeds up with more hash. Anyway, 4 MB is little, so one may think there is not enough eval cache on that, and it would be a reason why 32mb went faster, but on the other side it is an endgame position, and at least amyan evaluations there are very light so 4 mb is faster in nps. It depends on the heavyness of the eval.
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.