[FRIAM] Fwd: ABM

[Russell Standish]

On Sat, Jun 02, 2007 at 01:37:02PM -0600, Marcus G. Daniels wrote:
> Fortran 2003 seems like it could be a fine programming language for 
> agent models.   It has all the common OOP  features.  An argument for 
> Fortran, I suppose, would be superior numerical performance.   I doubt 
> it would be that big of a win, though, due to the fact that agent codes 
> tends to be rate limited by memory latency and CPU indirect mispredicts, 
> more than poor code generation of particular loops. 
> 

The legendry performance of Fortran is due to its restrictive
interpretation of arrays and variable in Fortran77 - all variables are
staticly allocated rather than stack allocated, arguments are never
aliased, everything has to be expressed as a loop over an array of
simple data types.

Come Fortran 90 or Fortran 2003 which adds a lot of interesting
features to the language that C++ or Java has, and performance is not
so easy to attain, unless one restricts oneself to the Fortran 77
subset.

It is certainly possible to achieve equivalent performance in C++ to
Fortran 77 code. However, it is not as easy to do so - it tends to
require a good understanding of C++ and how compilers optimise codes.

Cheers

-- 

----------------------------------------------------------------------------
A/Prof Russell Standish                  Phone 0425 253119 (mobile)
Mathematics                         	 
UNSW SYDNEY 2052         	         hpcoder at hpcoders.com.au
Australia                                http://www.hpcoders.com.au
----------------------------------------------------------------------------