Re: High resolution time routines

From: williamkempf_at_[hidden]
Date: 2001-03-22 15:36:38

--- In boost_at_y..., Toon Knapen <toon_at_s...> wrote:
> Beman Dawes wrote:
> > 
> > At 10:12 PM 3/21/2001, Jeremy Siek wrote:
> > 
> >  >I agree that a high resolution timer is needed. One quick way 
to provide
> >  >this is just to use the various high resolution timers that 
already
> >  >exist... wrapping them with a common interface. And the best 
part is that
> >  >someone's already done this (in C). The Interval Performance 
Monitoring
> >  >(IPM) package does this. It is part of the PHiPAC high-speed 
matrix
> >  >multiplication package. IPM is portable to most unix's. 
Boostifying this
> >  >package could be a good place to start.
> > 
> > The proposal that Bill Kempf pointed out
> > (http://www.cl.cam.ac.uk/~mgk25/c-time/) is of special interest 
because it
> > was put together by people active on the C language committee for 
possible
> > inclusion in a future C standard.
> > 
> > That doesn't mean Boost shouldn't look at other proposals - even 
the C
> > committee has looked at other proposals.  But in thinking about a
> > boostified  C++ interface, it might be nice to coordinate with 
the C folks
> > to avoid gratuitous conflicts.  I'll drop Marcus Kuhn an email to 
see if
> > interest exists.
> 
> Implementing the (future) C interface reusing the code/technology
> from IPM would make life easy, future safe and powerfull.

Yes, but we don't know what the "(future) C interface" is going to 
be.  This is what Beman wants to find out before going too far with 
this idea, I believe.

Bill Kempf