Author: Simon Waters
Date: 11:18:42 10/30/00
Go up one level in this thread
On October 29, 2000 at 04:55:39, Dan Newman wrote: >On October 28, 2000 at 15:13:01, Simon Waters wrote: > >>Anyone care to hazard a guess at what gnuchess 5 is trying to do with the Idepth >>variable and the DEPTH macro? >> >>I have read the source a few times, and pondered Aske's discussing of search >>algorithmns... but this leaves me cold. >> >>I spy a bit of systems decay I suspect, but I'd appreciate a few pointers. > >Gnuchess 5 is using the fractional ply trick. The DEPTH macro is >the number of fractional plies in one full ply, and Idepth is just >the nominal depth of the search iteration (in fractional plies). Thanks, you confirmed what I thought it was trying to do.... depth is how far it has got (in fractional ply, but always a multiple of DEPTH). Idepth is how far it is going in fractional ply - with extensions upto 2*Idepth. DEPTH is the number of fractional ply. What I missed was that Idepth is passed as depth to Score from ScoreRoot, and then depth as depth-1 setting up the recursion. And that the ChkCnt exists purely to stop the depth exceeding Idepth*2 when we have a sequence of checks (Perpetual maybe -- hmm). Having applied quite a few changes to the base code, my version is doing better on it's test cases than the latest code from Stuart, but I can't figure out exactly which change makes the benefits. I suspect that my "enhancements" don't help till I've moved all the bug fixes over.... and some of my compiler option changes seem to only work with my code changes argh.... Perhaps I should try a different test case and make sure it is not just a false positive.
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.