[Quantian-general] Load not being distributed continued...

[Dirk Eddelbuettel]

On 13 July 2007 at 15:12, David Hadley wrote:
| Sorry, I should have said this in the original post:
| 
| To say that there is no load distribution is not quite true. According
| to openMosixmigmon, one or two processes are being distributed but the
| processes that I start (e.g. R or folding at home) that are more CPU
| intensive do not even seem to appear in the openMosixmigmon, so I
| couldn't even manually drag the process to one of the clients.

Well, I guess youu are aware of the restriction plainly stated on the oM
FAQs?  No shared memory, and fairly strict limits on multithreaded apps. Are
you sure that folding at home passes these contraints?

My standard test is the 'BermudanSwaption' binary from the quantlib-examples
packages (that I maintain) which prices a few derivatives using Monte Carlo
-- that takes long enough to merit being passed around by openMosix yet not
too long for me to wait.  You may want to try that, or one of the simple Perl
examples from the oM docs or something similar.

Dirk

-- 
Hell, there are no rules here - we're trying to accomplish something. 
                                                  -- Thomas A. Edison