Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How to get more VPU time.

Author: José de Jesús García Ruvalcaba

Date: 09:15:33 04/10/99

Go up one level in this thread


On April 10, 1999 at 02:45:31, David Blackman wrote:

>I've downloaded it and had a quick look. There is probably one mistake:
>In search.c at line 182, in the function search(), where it says
>(nodes & 1000)
>i think this should be
>(nodes & 1023)
>or maybe
>(nodes % 1000)
>The idea here is he is trying to check the time once every thousand nodes, and
>the code he has doesn't do that.
>
>Another problem with this is nodes gets updated in qsearch() as well,
>so the time check code probably should be duplicated in there.
>
>I don't think this on its own should cause much slowness however. On Linux it
>certainly doesn't. I hope to get to a Sparc tomorrow long enough to give it a
>try.

	Now I know what was happening. My earlier reports were made based on blitz
games. Now I am playing at slower time controls, and Faile gets a full cpu for
itself. But it takes a few seconds after it is its turn to move (it does not
think on its opponent's time), and so in blitz games it hardly had time to get
over 40%.
	The version compiled with gcc still has problems, it does not get too much cpu
time even with slower time controls. Anyway I do not plan to use it when I have
a choice of using cc (:
	Thanks to you, to Manuel and to all that helped.



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.