Author: Miguel A. Ballicora
Date: 09:05:31 04/06/01
Go up one level in this thread
On April 05, 2001 at 08:56:14, Gian-Carlo Pascutto wrote: >On April 05, 2001 at 06:54:30, David Rasmussen wrote: > >>But have anybody tried a seperate Qsearch hashtable? The very fact that the >>Qsearch generates so many notes, and so much time is spent there, seems to >>dictate that the dynamic programming principle of a trans/ref table would be >>very beneficial here too. > >Sjeng uses a seperate Qsearch hashtable. > >It's a gain everytime I test it. Even with SEE, because even though >the hits are lower, the time saved is greater. > >-- >GCP I thought about this; however, I haven't tried yet. The question is whether the memory allocated for a separate Qsearch hash table could be used more efficiently for the normal table. In other words, what's better? 16 MB hashtables or 12 MB ht + 4 MB Qsearch hashtables? (or any combination). The good thing about qsearch tables is that I bet that It will have a high hitting ration between iterations (some qsearch lines should be exactly the same) and it would be more beneficial at faster time controls when the qsearch takes a lot of time. However, I wonder if at longer time controls become less useful. Regards, Miguel
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.