[Beowulf] A start in Parallel Programming?

Joe Landman landman at scalableinformatics.com
Tue Mar 13 11:41:23 PDT 2007


Jeffrey B. Layton wrote:

> Ouch this hurts! I've gotten the same rubbing from RGB. I hate to
> say it, but I'm a Fortran programmer ("Hi! My name is Jeff and I'm
> a Fortran programmer. Hi! Jeff!"). I know C reasonably well (except
> for the esoteric garbage) and I code in Python and the usual assortment
> of shell/interpreted language.

[begins preparations for the language war to follow]

You haven't lived until you typed in a program printed in a US 
government book from the 1960s, into your favorite editor, and compiled 
it ... only to find ... it worked .  Considering I did this in early 
1990 in grad school ... not bad.

Some folks deride (insert your most hated language here) and promote 
(insert your most beloved language here).  All that matters at the end 
of the day is that a) your code works, b) is fast, c) someone else can 
maintain it in 20 years when you are off relaxing on a beach ... 
studying solitons in nature .... yeah, thats the ticket ...

Sadly there aren't many languages that will get you there.  Lots of my 
old C still works.  Little of my old C++ works.  All of my old perl and 
python work.  Java, well, never mind.  For the most part, the fortran 
still works.

And that is scary, as I started playing with it in 1984 or so.

As for learning it now, sure, if you need it.  It compiles down quite 
nicely.  Optimization is quite good for it.  There are massive amounts 
of publicly available libraries for it.  C has been gaining in this 
respect for a while, but we are talking about decades of library 
building with Fortran in mind.  This sort of inertia is very hard to 
overcome.  C++ is a mixed bag: if you really want performance, your C++ 
is going to look an aweful lot like C (and Fortran) at the lowest levels 
... you likely won't be doing deep object data structure dereferences or 
method calls, or object factories, or ... in deeply nested loops.  That 
is if you care about performance you won't do this.



-- 

Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web  : http://www.scalableinformatics.com
phone: +1 734 786 8423
fax  : +1 734 786 8452 or +1 866 888 3112
cell : +1 734 612 4615




More information about the Beowulf mailing list