Ive noticed some programs go through the ply usually with just one pv showing on each ply e.g. 3,4,5,6,7,8,9 etc.. while other programs show pv many times at the same depth e.g. 3,3,4,4,5,5,6,6,7,7,7,8,8 etc..
What causes this? Is it actually the search or is it the method of reporting the pv?
pv and ply
Moderator: Ras
-
Edmund
- Posts: 670
- Joined: Mon Dec 03, 2007 3:01 pm
- Location: Barcelona, Spain
Re: pv and ply
This depends on whether you report the pv once the ply has finished or once the pv changes.
The latter can happen a couple of times per ply either when at the root-depth the current bestmove falls outside the aspiration window or another move fails high.
The latter can happen a couple of times per ply either when at the root-depth the current bestmove falls outside the aspiration window or another move fails high.
-
hgm
- Posts: 28390
- Joined: Fri Mar 10, 2006 10:06 am
- Location: Amsterdam
- Full name: H G Muller
Re: pv and ply
I recently switched Fairy-Max from reporting after every iteration to reporting on every improvement of alpha in the root. Mainly because the latter is more convenient for multi-PV mode: when yu print the PV's immediately after the search found them, you don't have to worry about storing them.
Main disadvantage is that you either have to report a duplicate of the best PV at the end of the iteration, just for knowing the iteration has finished, and how much time and nodes it took, or accept that this info is not printed.
Main disadvantage is that you either have to report a duplicate of the best PV at the end of the iteration, just for knowing the iteration has finished, and how much time and nodes it took, or accept that this info is not printed.