Joel Kolstad wrote:
"Roy Schestowitz" <newsgroups@schestowitz.com> wrote in message
cs23u4$171u$1@godfrey.mcc.ac.uk">news:cs23u4$171u$1@godfrey.mcc.ac.uk...
Yes, it is much slower, but that's the inevitable disadvantage of using
high-level programming languages. It's a trade-off between abstraction,
i.e. ease of development and the resulting speed.
In the case of Matlab, I'd say it's slow mainly because of the way it's
interpreted. C++ is arguably as 'high level' or even higher than Matlab
(C++ having far more powerful object-oriented features, but Matlab having
better handling of multiple arguments and return values, for instance...),
yet it's much faster than equivalent Matlab code.
At least on Windows platforms, there is support for OpennGL graphics,
wich should speed it up. Other than that, I don't know of any hardware
acceleration.
It is true that MATLAB is sometimes slower than a corresponding C/C++
program, but it very much depends on the implementation. If you use
matrix operations and built-in functions, rather than looping (like you
probably would in C), I'd say the MATLAB implementation will be both
faster and numerically more stable.
Also, there's the issue of simulation time vs. development time. I'd
rather sit and wait for the simulation for one day (eating donuts) than
hunting seg. faults in C ;)
Cheers,
/Johan
|
|
