I'm using Ivan D. Iavanov's implementation and the speedup was not satisfactory. I changed my solution to the OpenMP version CSDP and work fine!. I was unaware of the existance of that version (I forgot to read a bit little more the user manual, sorry!). <br>
<br>The algorithm that I'm implementing tries to solve nonlinear SDP problems from second order static-output feedback controllers. It's very similar to the SQP nonlinear programming algorithm, but applied to nonlinear semidefinite programming.<br>
<br>Thanks for the answer and comments.<br><br>Oscar<br><br><div class="gmail_quote">On Wed, Aug 19, 2009 at 10:05 PM, Brian Borchers <span dir="ltr"><<a href="mailto:borchers@nmt.edu">borchers@nmt.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br>
It isn't clear whether you're talking about Ivan D. Iavanov's distributed<br>
memory parallel version of CSDP running on a distributed memory Beowulf<br>
cluster or whether you're talking about the regular version of CSDP<br>
using OpenMP for parallelization on a shared memory multiprocessor.<br>
Which is it?<br>
<br>
You also haven't told us anything about the size and other aspects of<br>
the problems that you're trying to solve- getting good parallel speedups<br>
on very small problems is hard. It would help to know about the problems<br>
that you're solving.<br>
<br>
</blockquote></div><br>