Fwd: [Beowulf] A start in Parallel Programming?

Nathan Moore nmoore at winona.edu
Tue Mar 20 07:59:29 PDT 2007


> I can't believe the GSL people invented a new interface for the one
> numerical interface which is now universal. Not to mention that they
> ignored lots of faster, free libraries like ATLAS and FFTW... what's
> the point of reinventing the wheel badly?

The different interface was a little unconventional, but after  
spending an afternoon with the manual and some example programs it  
didn't seem like too much hassle.

You must admit that their coverage of mathematical functions is  
pretty comprehensive (what other FREE library does the same thing?   
Netlib might have all the parts, but nothing is as comprehensive.).   
The sheer variety of random number generators available impressed the  
heck out of me (especially after using drand48() for far too long).

In my mind (and I've only used gsl for graduate research, ~20 years  
of CPU time spread over a few months) the main advantage of GSL is  
the almost everything numerical that one might need is in the  
library.  Its also available via gnu for any processor type, so if  
you're scrounging for time and need to run your code on sgi, sun,  
linux, ibm, i386 linux, and alpha, (because that's what's available)  
and do your development work in osx, its no big deal to "make" with  
the vendor's compiler.  (for a little while I tried curating 6  
different makefiles to support all the vendor's different ideas about  
how the lapack sgetrf (L-U factorization) should be called (-lessl, - 
lsunmath, -llapack, -lcomplib, -lcxml, -lacml, - 
l_whatever_intel_called_it.....  I've tried to forget where all these  
different libraries are stored)

Nathan
- - - - - - - - - - - - - - - - - - - - - - -
Nathan Moore
Physics, Pasteur 152
Winona State University
nmoore at winona.edu
AIM:nmoorewsu
- - - - - - - - - - - - - - - - - - - - - - -







More information about the Beowulf mailing list