Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is there going to be a deep sjeng 1.5 for all the bsds ?

Author: Gian-Carlo Pascutto

Date: 09:37:53 08/22/03

Go up one level in this thread


On August 22, 2003 at 11:47:32, Steven Edwards wrote:

>On August 22, 2003 at 04:25:17, Gian-Carlo Pascutto wrote:
>
>>On August 21, 2003 at 17:36:30, ERIQ wrote:
>>
>>>or just darwin. Also has anyone try to get it to work in freebsd yet and if so
>>>how good is it at using two processors.
>>
>>As I understand it BSD has some compatibility layers for Linux, so maybe
>>the Linux version works.
>>
>>But I wouldn't count on it - Mac OS X didn't have the full functionality
>>I needed to get multiprocessor support to work, so I had to disable it
>>in the Mac version. I assume other BSD's have the same problem.
>
>What problem?

Buggy shared memory support. Another problem I had was that my spinlocking
code wouldn't compile - kinda strange since it came right out the Linux
kernel.

I gave up when it became clear the Mac OS X libc was buggy as well.

>I have multiple CPU support via POSIX threads in the CT toolkit and the
>identical source works on dual CPU machines running Linux AND dual CPU machines
>running Mac OS/X 10.2.6 OpenBSD.  I have only a single POSIX mutex for the
>whole toolkit and it is accessed only when a thread is created or destroyed.

1) It needs to run on Windows as well.
2) It needs to be able to lock several thousand times per second.

--
GCP



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.