win95 again

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

win95 again

Erich Neuwirth-2
i installed octave 2.0.13 in win95
and it works.

BUT

loading times for m files still are enourmous,
it takes minutes before such a function is executed.

i tead the hints about alising drives
and did a test.

i put median.m
in the direcory
c:\octace\bin
where the binary itself resides and
i set
LOADPATH="."

even under these conditions it takes minutes
before octave executes the function in the m file.

are there any known tricks to improve file loading performace?


--
Erich Neuwirth <[hidden email]>
Computer Supported Didactics Working Group, Univ. Vienna
Visit our SunSITE at http://sunsite.univie.ac.at




Reply | Threaded
Open this post in threaded view
|

Re: win95 again

Fletcher, John P
> From:          neuwirth  <[hidden email]>
> Subject:       win95 again

> i installed octave 2.0.13 in win95
> and it works.
>
> BUT
>
> loading times for m files still are enourmous,
> it takes minutes before such a function is executed.
 
> are there any known tricks to improve file loading performace?
>
>
Has any work been done to implement octave using mingw32 rather than
Cygnus win32, which from teh web sites should make it quicker?

John Fletcher
 
-------------------------------------------------------------------
Dr John P. Fletcher          Tel: (44) 121 359 3611 ext 4625
Department of Chemical Engineering and Applied Chemistry (CEAC),
Aston University,            Fax: (44) 121 359 4094
Aston Triangle,              Email: [hidden email]
BIRMINGHAM B4 7ET  U.K.  
-------------------------------------------------------------------
CEAC Web site http://www.ceac.aston.ac.uk/
-------------------------------------------------------------------


Reply | Threaded
Open this post in threaded view
|

Re: win95 again

John W. Eaton-6
On 11-Jun-1998, John Fletcher <[hidden email]> wrote:

| Has any work been done to implement octave using mingw32 rather than
| Cygnus win32, which from teh web sites should make it quicker?

I don't know of anyone who is working on doing this.  If you or anyone
else is interested in working on the project, please get in touch with
me.

Thanks,

jwe


Reply | Threaded
Open this post in threaded view
|

win95 again

John W. Eaton-6
In reply to this post by Erich Neuwirth-2
On  8-Jun-1998, neuwirth <[hidden email]> wrote:

| i installed octave 2.0.13 in win95
| and it works.
|
| BUT
|
| loading times for m files still are enourmous,
| it takes minutes before such a function is executed.
|
| i tead the hints about alising drives
| and did a test.
|
| i put median.m
| in the direcory
| c:\octace\bin
| where the binary itself resides and
| i set
| LOADPATH="."
|
| even under these conditions it takes minutes
| before octave executes the function in the m file.

I've only done a small amount of testing on Windows NT/95 systems but
I've never experienced problems like this.  I have no idea what the
trouble might be.  Can someone please try to debug this problem?

Thanks,

jwe


Reply | Threaded
Open this post in threaded view
|

Re: win95 again

Mumit Khan-2
In reply to this post by Fletcher, John P
"John Fletcher" <[hidden email]> writes:
> Has any work been done to implement octave using mingw32 rather than
> Cygnus win32, which from teh web sites should make it quicker?
>

To get Octave to build under mingw32, one would have to emulate the
subprocess management and signals, both of which are hard to do without
resorting to Win32 API (the C runtime is strictly ANSI with just a few
POSIX routines thrown in for good measure).

I went through this when I ported our simulation software, and ended
up using Mingw32 for the computational part and cygwin32 for the
simulation management module which I didn't feel like rewriting.
Essentially I have the same code except for a few emulation routines
that runs on Unix and Win32.

The reason I went with Mingw32 is strictly performance -- in some cases
10-15x speedups for some of the CPU and/or I/O intensive numerical codes.
I still have nightmares however trying to get around bugs in some of
Microsoft C runtime routines (CRTDLL and MSVCRT).

Regards,
Mumit