From: panxing@my-dejanews.com
Newsgroups: comp.parallel.mpi
Subject: Re: Why my MPICH/LAM under linux give low float point performance
Date: Thu, 27 Aug 1998 20:12:57 GMT
Organization: Deja News - The Leader in Internet Discussion
Message-Id: <6s4ek9$74b$1@nnrp1.dejanews.com>
References: <6s3nt4$8kd$1@nnrp1.dejanews.com>


In article <6s3nt4$8kd$1@nnrp1.dejanews.com>,
  panxing@my-dejanews.com wrote:

Sorry, there are a little bug in the program, the printf should not be
timed.
But I test again, even the printf is not timed, it is slow too.

almost 2seconds .vs. 0.85 seconds, (MPI .vs. normal C)
>
> The program is :
> #include "mpi.h"
> #include <stdio.h>
> #include <math.h>
>
> void main(argc,argv)
> int argc;
> char *argv[];
> {
>     int myid, numprocs,n;
>         register        double mye, x, y, e;
>      register long i,j;
>         double startwtime, endwtime;
>
>     MPI_Init(&argc,&argv);
>     MPI_Comm_size(MPI_COMM_WORLD,&numprocs);
>     MPI_Comm_rank(MPI_COMM_WORLD,&myid);
>
>         x=1.0; y=1.00000001;
>         if (myid == 0)
>         {
>             startwtime = MPI_Wtime();
>         }
>             for (i =0;i<100000000;i++ )
>             {
>                 x =x*y;
>             }
>             mye = x;
>             if (myid == 0)
>             {
>                 printf("The my final e is %lf\n", mye);
>                 endwtime = MPI_Wtime();
>                 printf("wall clock time = %f\n",
>                        endwtime-startwtime);
>             }
>
>     MPI_Finalize();
> }
>
> -----== Posted via Deja News, The Leader in Internet Discussion ==-----
> http://www.dejanews.com/rg_mkgrp.xp   Create Your Own Free Member Forum
>

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/rg_mkgrp.xp   Create Your Own Free Member Forum