Q1. Where can I find the following details about the behavior of the
parallel search engine?
Suppose I execute a search with 6 machines, each with one processor.
(They could have 2 processors, but I'm trying to keep it simple.)
Q2. What determines/controls the load balancing between each of the 6
processors?
Q3. What determines how much of a given machine's CPU and memory the Oz
processes are allowed
to use?
Q4. Are any of the above configurable? And if so, how?
I'm in the process of working on two parallelized searches:
One hosted in a lab I control with 30 machines, and
One potentially hosted on a slice of the PlanetLab network.
In the latter case, any processes that run on that network have to be
polite,
and not potentially choke off other researcher's processes that may be
using the same host.
I should mention that before I actually try anything like 30 machines, I
plan to try just 2.
If anyone knows of any issues, or gotchas with this idea, please let me
know.
I'm still at the stage of converting my existing code to a functor, and
next I'm going to figure out how to enable
remote execution, as described in the Mozart documentation, so I haven't
tried any of this yet.
--------------------
George Rudolph
Assistant Professor of Computer Science
Thompson Hall 225
Math & Computer Science Dept.
The Citadel
171 Moultrie St.
Charleston, SC 29409
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
