Re: [Boost-users] date_time clocks, microsec_clock, etc

From: Jeff Garland (jeff_at_[hidden])
Date: 2004-05-06 08:46:28

On Thu, 06 May 2004 02:46:30 -0500, Aaron W. LaFramboise wrote
> I have prototype implementations for a few simple Windows clocks I 
> have in mind in works.  I'll post when I get a chance to finish something.

Great.  BTW, did you see the earlier thread on the boost::timer?  There are
some pointers to it on the Wiki in case you missed it:

http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl?GDTL/Timer

> In the meantime, perhaps this might be interesting to someone.  I 
> wrote this two days ago to assess the capabilities of the clocks on 
> Windows to determine what might be the most suitable as a time 
> source.  It outputs a combination of hardcoded information about the 
> documented capabilities of clocks, plus information it has 
> determined real-time about the capabilities of clocks from the OS. 
>  It should be standard C++ with documented Windows extensions.  I 
> tested on GCC, but it would probably work for any Windows compiler 
> with little or no modifications.  The rdtsc support will only work 
> on GCC, though.
> 
> Note that the implementation is very naïve, and there are probably 
> numerous flaws.  Its the best I could come up with without wasting a 
> whole lot of time.  I am certainly interested in possible 
> improvements.  It takes about 30 seconds to run.
> 
> On my platform (Windows XP Professional, GCC 3.3.2, MSVCRT runtime, 
> Intel Pentium 4 1.4GHz), it has the following output.  I don't have 
> access to a whole lot of other machines presently, so if you have a 
> Windows computer where these tests show interesting results, I would 
> certainly be interested in privately seeing the output on your machine.

Very interesting.  I'll play around with this.  If the last 2 measurements are
correct they would very precise clocks for timing.

Jeff