Re: [MLUG] mono

[Michael <EMAIL:PROTECTED>]

>    FWIW, a simple for loop is about 300 times faster
>    in mono than in Octave.
>
> I would definitely like to see more interoperability between Octave, 
> R, some of the graphics programs, symbolic math programs and 
> typesetting programs.

Sadly, this doesn't mean that Octave.NET would be 300 times faster than 
Octave. It means that the .NET virtual machine and whatever compiler 
produced the .NET program together produce a loop about 300 times faster 
than the existing Octave parser and excution core. When ported to .NET 
Octave could add it's own slowness to whatever inherit slowness .NET has 
to together create something even slower. Hopefully the porting will 
result in code cleanup and refactoring that will allow Octave programs 
to run at close to that 300x speed though. A common execution core 
shared between languages is a good idea partly because the reuse of code 
should result in a better and faster execution core. From this post it 
sounds as if Octave isn't terribly effecient in how it does this right now.

More interesting comparisons would be comparing the same code done in C 
, Python, Java, and C#. Comparisons between CPython (C), IronPython 
(.NET), Python with Psyco/PyPy (C with JIT), and JPython (Java) enabled 
would interest me too since they'd more greatly explore speed 
relationships of the same language in different execution cores.

--
Michael <EMAIL:PROTECTED>
http://kavlon.org


_______________________________________________
members mailing list
EMAIL:PROTECTED
http://mlug.missouri.edu/mailman/listinfo/members