Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 0x88 and move generator speed

Author: Miguel A. Ballicora

Date: 14:10:13 01/29/01

Go up one level in this thread


On January 29, 2001 at 16:14:52, José Carlos wrote:

>On January 29, 2001 at 14:37:45, Dann Corbit wrote:
>
>>On January 29, 2001 at 09:08:59, José Carlos wrote:
>
>[snip]
>
>>For something more interesting, try both Vincent's position and also this one:
[D]8/8/p1r5/6k1/KP6/8/8/5R2 b - -
>>
>>with both board formats.
>
>  What should I do with this position? Generate moves? I mean, what's the test?
>If it is the test about tablebases, I'm rewriting the engine from scratch, no
>tablebases yet... :(
>
>  José C.

Hola Jose,

I guess that the idea would be to test the move generator in both, "dense" and
"sparse" positions.

BTW, one test I would like to see is actually a "perft like" test (as in
Crafty), where the whole tree is visited, generating and make/unmaking the moves
and tested for legality at some point. It is like the engine with no evaluation
whatsoever. Can you guys do something like that to compare?

I am using bitboards (which is supposed to be ok for generating) but the
bottleneck is in make/unmake. My program is terribly slow in this respect.
If I do perft testing the nps, I get 85 knps in a AMD K6-2 400 mhz. Awful!
When I start adding things... I end up in 30-40 knps.
I consider the possibility to use 0x88, but... once you use bitboards,
everything is related to that... I will have to rewrite everything.

Regards,
Miguel



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.