I have a moderately complex fortran program, compiled with Compaq visual
fortran. This is run from the command line.
 If I run this with output to the screen it runs in a few seconds and gets
almost 100% of the CPU.
 If output is to a file (opened in the program) then it takes minutes to run
getting between 0 - 5 % of the CPU.
Running with output to stdout but redirected to a file (>) is also slow but
faster than when the file is opened in the program (different buffer sizes
?)

I have tried playing around with system settings for foreground and
background without changing this behaviour.

has anyony any suggestions as to whatt is causing this and how to fix it?

Steve

"Stephen White" <S.White@irl.cri.nz> wrote:

>I have a moderately complex fortran program, compiled with Compaq visual
>fortran. This is run from the command line.
> If I run this with output to the screen it runs in a few seconds and gets
>almost 100% of the CPU.
> If output is to a file (opened in the program) then it takes minutes to run
>getting between 0 - 5 % of the CPU.
>Running with output to stdout but redirected to a file (>) is also slow but
>faster than when the file is opened in the program (different buffer sizes
>?)
>
>I have tried playing around with system settings for foreground and
>background without changing this behaviour.
>
>has anyony any suggestions as to whatt is causing this and how to fix it?
>
>Steve
>
>
Have you tried changing (increasing) the block size for the
external file?

>


-- 
Mike Prager, NOAA, Beaufort, NC
Address spam-trapped; remove color to reply.
* Opinions expressed are personal and not represented otherwise.
* Any use of tradenames does not constitute a NOAA endorsement.

"Stephen White" <S.White@irl.cri.nz> wrote in message
news:c7pvrs$30gv$1@news.comnet.co.nz...
> I have a moderately complex fortran program, compiled with Compaq visual
> fortran. This is run from the command line.
>  If I run this with output to the screen it runs in a few seconds and gets
> almost 100% of the CPU.
>  If output is to a file (opened in the program) then it takes minutes to
run
> getting between 0 - 5 % of the CPU.
> Running with output to stdout but redirected to a file (>) is also slow
but
> faster than when the file is opened in the program (different buffer sizes
> ?)
>
> I have tried playing around with system settings for foreground and
> background without changing this behaviour.
>
> has anyony any suggestions as to whatt is causing this and how to fix it?
>
> Steve
>
Some of the usual things to check
Virus checker settings
fragmentation or disk errors, or disk 90% full
problems with path setting
file system type, or unusually small cluster size set

>From: "Stephen White" S.White@irl.cri.nz 
>Date: 5/11/04 12:37 AM Pacific Daylight Time
>Message-id: <c7pvrs$30gv$1@news.comnet.co.nz>
>
>I have a moderately complex fortran program, compiled with Compaq visual
>fortran. This is run from the command line.
> If I run this with output to the screen it runs in a few seconds and gets
>almost 100% of the CPU.
> If output is to a file (opened in the program) then it takes minutes to run
>getting between 0 - 5 % of the CPU.
>Running with output to stdout but redirected to a file (>) is also slow but
>faster than when the file is opened in the program (different buffer sizes
>?)
>
>I have tried playing around with system settings for foreground and
>background without changing this behaviour.
>
>has anyony any suggestions as to whatt is causing this and how to fix it?
>
>Steve
>

It sounds like you have something much more going on than a program simply
running slightly slower under a different set of circumstances.   I'm betting
that you have "other" misbehaving software on your computer system!!!   I'd
guess either network related software or virus/firewall software.  Of course...
that's merely guesswork.

I had the opposite problem with an NT system.  Running simple DOS box programs
took forever to get off the ground.   It ended up that some of the network
software on our system was misbehaving and mucking up the works for any program
that required the DOS command processor.

I'm still not sure why it took forever for me to convince our IT
"professionals" that the apparently slow running code wasn't because "DOS is
obsolete".  They tried really hard to convince me that our "engineering"
software should either be investigated for bugs or be updated to "Windows".  
Oh well,  that's what happens when those with the only keys to Administrative
Rights couldn't write a "Hello World" program to save their lives.

Dan  :-)

Are you writing to a local disk, or to a network (mapped) drive?

Is it a one chunk of data, or multiple small writes to a network drive?

Our experience is that WinXP network drives are not intended for heavy traffic.
Our programs nearly stop to a halt when writing to a network drive; otherwise,
the writing time is negligible. It has happened locally, and on customers
computers. Maybe our fault, but it does not happen with NT/2000..

Good luck,
Juan Carlos

Stephen White wrote:

> I have a moderately complex fortran program, compiled with Compaq visual
> fortran. This is run from the command line.
>  If I run this with output to the screen it runs in a few seconds and gets
> almost 100% of the CPU.
>  If output is to a file (opened in the program) then it takes minutes to run
> getting between 0 - 5 % of the CPU.
> Running with output to stdout but redirected to a file (>) is also slow but
> faster than when the file is opened in the program (different buffer sizes
> ?)
>
> I have tried playing around with system settings for foreground and
> background without changing this behaviour.
>
> has anyony any suggestions as to whatt is causing this and how to fix it?
>
> Steve