Author: Dann Corbit
Date: 10:15:09 10/26/99
Go up one level in this thread
On October 25, 1999 at 18:21:20, Michel Langeveld wrote: [snip] >This Is what I thought also but I can remember from my latest profiler results >that evaluate was _not_ that most costing function. I shall recheck this... > >Btw. (Same other point in the source) The book.c source is awfull... The only >good thing about it now is that is is tested on multi platforms and that is >works.. But I'm sure generating a book can be done in a more smarter form then >generating all kind of tempory files. I hate also the giant malloc on line 864. Book only runs during the creation of the book (for me, about once every 2 months or so). I think trying to make the code cleaner or faster is not such a great idea, because it has very little effect on the execution of the program. If you are modifying the current book code, you might want to hold off a bit, since things are going to change quite a bit for version 17 of crafty. > >Maybe a inserting with using a arching the book binary can clean much of the >code and make it easily understandable. Since the makefile.nt works perfect for >my I'm trying to changing things now in a good way at the moment... I shall give >you all the modifications when the appear to better/faster/etc. > >I'm almost ready with stuffing all book positions to an epd file. So when >generating a book simultaneasly and epd file is created of all positions in the >book. This was only maybe 10 lines of adding code... I would like to see the effect of adding *ALL* the CAP data as book positions. There would be no won/loss/draw statistics, but it would be very valuable (especially in blitz games) whenever there is a hit. It would also be very nice to get the missing positions, ordered by frequency of occurence, in order to run them through C.A.P. and fill out the book 100%.
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.