Prev: Calling .NET from MATLAB
Next: Image Averaging
From: Rune Allnor on 14 Sep 2006 19:29

Steve Amphlett skrev:
> Rune Allnor wrote:
> >
> > The issue that REALLY killed my interest in CS was the MSDOS
> > memory mapping madness. In -90 I spent the equivalent of $3000 on
> > a once-in-a-lifetime offer: 386 processor (prep'ed for the 387,
> > which
> > I installed a year later), 40 MB hard disk and a whopping 2 MB RAM.
> >
> > Yeah. Truly a once-in-a-lifetime offer, all right...
> >
> > Anyway, at the time I knew my way reasonably well around
> > x86 assembler code. But there was no way I was able to get
> > out of that 64 kbyte page size inherent in the MSDOS memory
> > schema; let alone get access to that elusive extra 1 MB of RAM
> > I had paid deer $$ for. When I realized that actually getting
> > access to the available HW resources would require some 90%
> > of the coding effort in any realistic project, I started looking
> > for other things to do.
>
> Ok, my second big assembler/machine code project was using an Acorn
> ARM chip (~1988-1989). That was true 32-bit , woooo! An interest
> reawakened. Mind you, x86/x86_64/EM64T/IA64/etc these days, who
> cares? I wonder who actually writes assembler for these chips and
> how long their skills are useable.

Nowadays assembler is mostly used in high performance/constrained
resource applications, like all these small wireless gadgets that
are all over the place.

It is 15+ years since I last wrote any assembler code. Nevertheless,
the principles are still valid. Or maybe not. I got slightly depressed
when I discovered that a C++ compiler produced executables that
ran twice as fast from "naive" source code compared to the code I
had emploied all my skills, albeit rusty, to optimize.

Rune

First  |  Prev  | 
Pages: 1 2 3 4 5 6
Prev: Calling .NET from MATLAB
Next: Image Averaging