🚀 go-pugleaf

davis@adrs1.dseg.ti.com (Tom L. Davis) wrote:
>
> In article <9602120100.AA23318@ues5.cern.ch>, Dan Pop <Dan.Pop@cern.ch> wrote:
> 
>     davis@adrs1.dseg.ti.com (Tom L. Davis) writes:
>     
>     >Actually, the best way to do scientific or engineering problems is to use
>     >commerical apps like Matlab. Your productivity will increase by about a
>     >factor of 10. 
>     
>     One of the best jokes I read in this newsgroup for a long time.
> 
> It wasn't meant as a joke, Don. Do you even know what Matlab is? As an
> engineer who has written thousands of lines of Fortran (long ago) and C in
> pursuit of answers to engineering problems, I assure you that Matlab is
> far superior. 
>    
>     Try to ask the people who simulate particle physics experiments or
>     process seismical data or do any other kind of serious scientific or 
>     engineering programming what do they think about the productivity increase
>     factor of Matlab for their work.
> 

Just thought I'd enter ny 2 cents on this topic. I am
and have been an extensive user of matlab for close to 10
yars now, primarily for simulation of systems and 
development of control laws, as well as studies of
simulated annealing. Matlab has many good points
and makes development of code an order of magnitude
faster than traditional languages like C and Fortran.
However it is an interpreted language and typically
runs an order of magnitude slower than C or Fortran.
I've read in Matlab newsgroup that if one can vectorize
the problem enough the performance is not so bad, but in
my personal experience with simulation of control systems
it is never less than an order of magnitde slower. So each of
the above posts has some validity, but 
the idea that Matlab is the panacea for all problems or that it is
only useful for tou problems is a gross simplification. It does
somethings well and somethings poorly. Supposedly with the new
Matlab compiler and the simulink accelerator Mathworks is
addressing the performance drawback. (Personally I wouldn't hold
my breath. Since releasing version 4 Mathworks seems hell bent on
destroying a great program. My expericne with the simulink
accelerator is that programs whose numerical integration
is sensitive in regular mode are unstable in accelerated mode.)

In sumary I guess I wished to say that net contributors
should be less general in there comments. Never say never:)
as the case would be. 




>     Dan
>     -- 
>     Dan Pop
>     CERN, CN Division
>     Email: danpop@mail.cern.ch 
>     Mail:  CERN - PPE, Bat. 31 R-004, CH-1211 Geneve 23, Switzerland
> 
> 
> What on earth warrants this kind of post? Does anybody understand the
> workings of Mr. Pop's mind? I can assure you that I don't. 
> 
> 
> Tom

Lyndon