Author: Eugene Nalimov
Date: 15:41:28 08/21/98
Go up one level in this thread
On August 21, 1998 at 18:03:55, Larry Coon wrote: >On August 21, 1998 at 16:27:09, Robert Hyatt wrote: > >>I would suggest (a) looking at the "epdkit" source from Steven Edwards, as >>it has code to probe the database... and (b) finding the source for tbgen, >>the program he wrote to build these things, which might help... > >Thanks for the reply. I found the epd code, as you suggested. >Frankly, I was hoping to avoid using the lookup code if I could, >since specs are easier to work with and algorithms often have >subtelty that's easy to miss. But if I have to, I have to.... > >(I want to use some of the tb information in a completely >different environment, so just linking to the existing code >isn't possible.) > >I haven't found tbgen yet, but I haven't looked very hard yet >either. I hope that next week I'll release alpha release of my own tablebase generator. It can generate tablebases in 2 formats: old (compatible with SJE) and new (more compact; en passant capture implemented). Design principles: 1. Generator must be readable. 2. I prefer to use RAM instead of disk; RAM is cheap. 3. I prefer not to use disk, if not necessary; so, sometimes I re-generate necessary information. Reason for that is that CPUs become faster much faster than disks. 4. Tablebases must be compact. 5. I do not compress files yet, just use better indexing scheme. Compressing is next project. Eugene
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.