Re: [pooma-dev] Patch: Doof2d: Eliminate Duplicate Computations

[Jeffrey Oldham]

On Fri, Oct 12, 2001 at 02:58:26PM -0600, James Crotinger wrote:
> I wouldn't think you'd want to time the initial condition setup.

It's my understanding that the benchmark class subtracts out the
runSetup() time after calling run().

> > -----Original Message-----
> > From: Jeffrey Oldham [mailto:address@hidden
> > Sent: Friday, October 12, 2001 2:51 PM
> > To: address@hidden
> > Subject: [pooma-dev] Patch: Doof2d: Eliminate Duplicate Computations
> > 
> > 
> > 2001-10-12  Jeffrey D. Oldham  <address@hidden>
> > 
> >         This patch reduces duplicate computation in the 
> > Doof2d benchmarks.
> >         Both initialize() and run() set the initial conditions.  In
> >         consultations with Stephen Smith, we have the untimed 
> > initialize()
> >         ensure that all the allocated memory pages are 
> > actually created
> >         and have run() set up the initial conditions.
> > 
> >         * Doof2d.h (Doof2dBase::initialize): Replace initialize() with
> >         statements to ensure all memory pages are actually created.
> >         (Doof2dBase::setInitialConditions): Removed in favor 
> > of code in
> >         initialize().
> >         * Doof2dInC.h (Doof2dInC::initialize) and
> >         (Doof2dInC::setInitialConditions): Likewise.
> >         (Doof2dCoefsInC::initialize) and
> >         (Doof2dCoefsInC::setInitialConditions): Likewise.
> > 
> > Tested on       sequential Linux using g++3.1 by running Doof2d
> > Approved by     Stephen Smith
> > Applied to      mainline

Thanks,
Jeffrey D. Oldham
address@hidden