C++ programming (was Newbie Alert: Beginning parallel program ming with Scyld)

Eray Ozkural erayo at cs.bilkent.edu.tr
Sun Oct 20 04:10:17 PDT 2002


On Saturday 19 October 2002 03:01, Gerry Creager N5JXS wrote:
> Now, I must take exception the fine elcture on "What is a Library."
> I'll have you know *I* was using reusable code back in the days of 029
> keypunches, and IBM's IEBPRINTPUNCH utilities.  I called 'em subroutines
> and functions.  I called them directly or indirectly, and I reused them
> by loading them into the hopper every time I ran the job.  I lived for
> the times when I could justify a data _TAPE_ instead of 3 boxes of data
> _CARDS_, though!
>
> Winkingly yours, Gerry
>
> And for the record... How many REALLY recall the "Do Inspite Of"
> construct in the Fortran77 draft standards?

This has got be one of the coolest posts ever on this list :)

Still, there is a moot point. The hard fact is that there is no parallel 
programming language despite all the claims. In my opinion the only one that 
comes close is HPF and a few of the functional PL compilers (like ghc), but I 
never had the leisure to try them so that is assuming you trust the papers.

My programming language of choice for parallel programming is C++, since it is 
a low level language (in spite of what you may think) but still suffficiently 
abstract to allow me to code in a rather modular and comprehensible way. C++ 
is eligible for the role of assembly in supercomputing. With C++/MPI, I can 
write generic communication routines, object-based data structure libs and 
such, none of which is really possible in the two remaining culprits.

My order of preference for a parallel programming project would be
1) C++
2) FORTRAN
3) C

This is the best we can do with three languages that all suck.

I had been planning to write a parallel programming language that would have a 
nice functional semantics and efficient execution during my msc. thesis. I 
was thinking, maybe I could find a program representation through which an 
automatic code/data distribution/replication could be derived (overly 
simplifying things) What I found was that the thesis exploded in my hand and 
I had to find another subject :) If somebody offers me an inordinate amount 
of resources, I can get back to that subject gladly. I guarantee no results 
however ;)

Cheers,

PS: I think I truly hated C when I coded a parallel quicksort routine for a 
lamer at our department (I can give you his name so you can avoid him in the 
future). Besides, he didn't pay me the complete amount of money he promised.

-- 
Eray Ozkural (exa) <erayo at cs.bilkent.edu.tr>
Comp. Sci. Dept., Bilkent University, Ankara
www: http://www.cs.bilkent.edu.tr/~erayo  Malfunction: http://mp3.com/ariza
GPG public key fingerprint: 360C 852F 88B0 A745 F31B  EA0F 7C07 AE16 874D 539C




More information about the Beowulf mailing list