| Re: call_external -> multiprocessing? [message #12947] |
Wed, 23 September 1998 00:00 |
steinhh
Messages: 260
Registered: June 1994
|
Senior Member |
|
|
In article <3607EAF9.CFC778DA@casa.colorado.edu>
Remy Indebetouw <indebeto@casa.colorado.edu> writes:
> I run on a 10 processor SGI Onyx and I want to have C code run on all
> the processors (parallelized using APO options) when called with
> call_external. It seems that call-external only uses a single processor
> because
> it still runs under the IDL process (I know next to nothing about
> dynamically linked libraries).
>
> Anyone know the answer?
No, but I suspect it may be possible to do this (having
a few routines using e.g., threads to parallelize stuff,
while the "main program" knows nothing about it. This
depends a lot on what "parallelized using APO options"
means, though! (Totally unknown to me...)
Why not just try it? Write some extremely time consuming,
easily parallelizable (ugh, what a word) routine, try to
compile with the APO options (?) into a shareable
library, and see what kind of (CPU) resources the program
uses, and what the execution time is with/without the
APO options?
Regards,
Stein Vidar
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
