Author: Anthony Cozzie
Date: 15:19:59 06/15/04
Go up one level in this thread
On June 15, 2004 at 17:24:59, Eric Oldre wrote: > >First off, I've taken everyone's advice from last week and changed my engine so >that most functions take a pointer to a board object as a parameter instead of >operating on a global board. > >I now have it doing a basic alpha-beta search with QSearch, a basic evaluation, >and MMV/LVA move ordering. It is also returning and printing out the PV. > >At this point (having hardly anything tuned) it's already as fast as my old >engine in vb.net. Although as I improve the evaluation it will slow down of >course. > >My next step is to get it implementing the winboard protocol, perhaps UCI as >well. I've decided to take a approach where i have the main thread read commands >from winboard, and a separate "thinker" thread work on the problem. similar in >concept to what i saw in the Gerbil source. > >Since this is the first program I've written in C. I don't have much experience >with multi-threading in C. I've started to create something using the >CreateThread and ExitThread calls in <windows.h>. I was wondering if people felt >that was a good idea, or if I'd be better off trying to learn about POSIX >threading. > >Is there a 3rd approach to look at? > >Is there a performance advantage to either? > >Does one have a much easier learning curve than the other? What i've done so far >with the <windows.h> stuff hasn't seemed to difficult to grasp, but I'm not >terribly far along yet. > >Thanks, >Eric Oldre My advice: just poll every so often. Threaded code is _tough_ to debug. I've had 4 different IO thread implementations in Zappa to date. I'm 99% sure I have it down now, but it really wasn't worth it. Humans are simply bad at parallel programming. anthony
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.