Do you guys clear out the "new" entries in the history table if you fail your aspiration window check?
Are there any disadvantages to leaving the updated history values in on the second/third call to the search?
Currently I'm reseting them, but that code is kinda slow on very low depths, on high depths of course its a negligable amount of time.
History Heuristic and Aspiration Windows
Moderator: Ras
-
cdani
- Posts: 2204
- Joined: Sat Jan 18, 2014 10:24 am
- Location: Andorra
Re: History Heuristic and Aspiration Windows
Not on Andscacs. I clear it only at the start of the search.
Also, and this is specific for Andscacs, I found that is good to clear it when the result of the iteration is 0.00.
Also, and this is specific for Andscacs, I found that is good to clear it when the result of the iteration is 0.00.
Daniel José -
http://www.andscacs.com
-
hgm
- Posts: 28480
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: History Heuristic and Aspiration Windows
Would it be better to clear, or just divide the values by 2? I would imagine that positions do not change so much from one move to the next that the information suddenly becomes useless.
-
Fabio Gobbato
- Posts: 219
- Joined: Fri Apr 11, 2014 10:45 am
- Full name: Fabio Gobbato
Re: History Heuristic and Aspiration Windows
I clear the history table at the beginning of every iteration with a little gain respect to not clear the table.
But if the aspiration window fail I keep the same table. I haven't tried to clear it.
But if the aspiration window fail I keep the same table. I haven't tried to clear it.