Mail Archives: djgpp/1997/11/24/10:07:15
Thomas Knudsen <tk AT geb DOT gfy DOT ku DOT dk> wrote:
> On 22 Nov 1997, Ed Leiby wrote:
>
> ...
> > However, I have recently been hacking around w/ a DSP simulation
> > compiled under gcc ( rev 2.7.2.1 ) using rhide ( rev 1.3 ) inside a
> > win95 dos box on a Pentium Pro 200Mhz machine. I had previously been
> > using Borland 5.01 set for a win95 console app, but had not been using
> > any win32 specific anything. Well, the app compiled like a dream under
> > gcc and ran right off, even reproducing results obtained under Borland.
> > The problem is that it seems to run at two different speeds: As fast as
> > Borland, and 2-3 times slower than Borland. I have tried using
> > cwsdpmi,cwsdpr0 (rev 0.90+) and win95 DPMI server. However, to date I
> > have not been able to correlate any action on my part or setup of the PC
> > to the different execution speeds. I thought for a while that it had to
> ...
> >
> > Ed Leiby
>
>
> I have had the same experience with a GRX2.2+DJGPP based graphics program,
> running in a DOS box under Win95. The symptoms are quite hard to describe
> consistently (as they are quite inconsistent ;-), but at least *most* of
> the time, the program tends to run slow, when running by itself, i.e.
> only sharing the processor power with the Win95 MegaKludge (TM). However,
> if I open *another* DOS box, and have it sitting idle in the background,
> suddenly the program runs at more than double speed. As it's a program
> doing scientific animations (visual number crunching?) this really makes a
> discernible difference---it would just be so nice to understand the cause
> of the strange behaviour: Optimising the code for speed is pretty hard
> when run times are inconsistent. So if any OS gurus could make sense from
> this, please don't hesitate...
>
> Perhaps, I should add, that *sometimes*, I have to open 2 more DOS boxes
> to get the speed increase...
That's the third user reporting a similar thing (the first was the one that
says your programs runs faster from inside of RHIDE). Seems that W95 scheduler
have some strnage behavior with DJGPP applications.
Does anybody have an idea on how to meassure it or figure out what exactly is
the problem?
SET
------------------------------------ 0 --------------------------------
Visit my home page: http://www.geocities.com/SiliconValley/Vista/6552/
Salvador Eduardo Tropea (SET). (Electronics Engineer)
Alternative e-mail: set-sot AT usa DOT net - ICQ: 2951574
Address: Curapaligue 2124, Caseros, 3 de Febrero
Buenos Aires, (1678), ARGENTINA
TE: +(541) 759 0013
- Raw text -