Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty and single-computer winboard matches

Author: Dave Gomboc

Date: 10:55:25 10/09/99

Go up one level in this thread


On October 09, 1999 at 13:02:38, Christophe Theron wrote:

>On October 09, 1999 at 07:30:48, Vincent Lejeune wrote:
>
>>On October 08, 1999 at 16:43:37, Christophe Theron wrote:
>>
>>>On October 08, 1999 at 06:14:23, Vincent Lejeune wrote:
>>>
>>>>On October 08, 1999 at 04:16:57, Christophe Theron wrote:
>>>>
>>>>>On October 08, 1999 at 03:42:54, Dave Gomboc wrote:
>>>>>
>>>>>>On October 07, 1999 at 23:29:05, Christophe Theron wrote:
>>>>>>
>>>>>>>That's a real problem on PC.
>>>>>>>
>>>>>>>The timer clicks 65536 times in 1 hour, which makes something close to 18.2
>>>>>>>times per second, or a 0.05s timer resolution.
>>>>>>>
>>>>>>>This numbers come from the prehistoric IBM PC 4.77MHz and have never been
>>>>>>>changed in 20 years, for compatibility reasons.
>>>>>>>
>>>>>>>Even Windows programmers did not dare to change this. You have time functions in
>>>>>>>Windows, they returns values in milliseconds, but still the resolution is about
>>>>>>>0.05s!!!
>>>>>>
>>>>>>Use performance tick counts if you develop for Win32.  (Actually, I am not 100%
>>>>>>sure that Win98 supports them. :-(  WinNT does for sure.)  Their resolution is
>>>>>>very good.
>>>>>>
>>>>>>Dave
>>>>>
>>>>>Not sure this works. In the experiences I have done, the timer resolution was
>>>>>always 0.05s. I'm pressimistic about getting something more accurate under
>>>>>Windows 9x.
>>>>>
>>>>>This is not a big problem anyway...
>>>>>
>>>>>
>>>>>    Christophe
>>>>
>>>>I've tested the gettickcount() Win32 api, it's documented as have a 1
>>>>millisecond time resolution and it have it effectivly.
>>>>
>>>>The win32 functions are usable in all Windows programming language (C++ (Borland
>>>>and Microsoft), Delphi, ...)
>>>
>>>
>>>Good to know. Thanks.
>>>
>>>
>>>    Christophe
>>
>>the gettickcount() function have one little drawback: it's a 32 bits variable.
>>So every (about) 47 days, the counter going back to 0, you need to add simple
>>test if the gettickcount()-starttime is < 0.
>
>The (hand written) function I use currently has exactly the same problem, so it
>is not a killer for me.
>
>
>    Christophe

Note that this bug has existed since 1995, and took more than three years to
discover.  Apparently, machines that stay up for 47 days are not too common in
the Windows world. :-)

Dave



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.