- Home - Download
- Run - Report Result -
Download
-
download the archive:
-
unpack the archive in a new directory and make sure this directory
appears in your CLASSPATH. These classes have to be accessible from
each machine involved in the benchmark (the central dispatcher as well
as all the machines running rendering engines).
Run
-
Choose a set of URLs for the rendering engines to bind to and make
a text file out of it. Save this file under any name with extension .hosts,
like local.hosts, mycluster.hosts, etc... Here is a sample
for this file:
# This file lists the urls where
# the dispatcher can find the rendering engines
//ll.inria.fr/vm1
//ll.inria.fr/vm2
//beaupre.inria.fr/vm1
//alaleur.inria.fr/vm1
Note that we could also choose to run all rendering engines on the same
host.
-
start rmiregistry and a rendering engine on every machine involved in the benchmark (rendering
engines + dispatcher).
user@ll> java fr.inria.proactive.Node //ll.inria.fr/vm1
user@ll> java fr.inria.proactive.Node //ll.inria.fr/vm2
user@beaupre> java fr.inria.proactive.Node //beaupre.inria.fr/vm1
user@alaleur> java fr.inria.proactive.Node //alaleur.inria.fr/vm1
-
start the central
dispatcher with a command-line like
user@tuba> java fr.inria.proactive.examples.c3d.C3DDispatcher mycluster.hosts &
A window pops up that looks like this:
Then start the benchark with
user@tuba> java fr.inria.proactive.examples.c3d.C3DUser&
A window pops up that looks like this:
-
After having rendered 4 views of the scene, this window disappears and
a log file is generated. This file is named benchX.txt, where X is the
number of rendering engines involved. The most important figure in that
file is the score, which gives an indication of the performance of the
platform when it comes to serialization and RMI. This figure is normalized
to a Pentium 200 PC with 128Mbytes of RAM running Windows NT 4.0, where
the dispatcher, the two rendering engines and the user frame where running
on the local host.
Report Results
You can report your results by email to Julien
Vayssière. Please do not forget to include:
-
A description of the machines involved in the benchmark: hardware, operating
system and which JVM and RMI/Serialization implementation you're using.
-
The score this platform achieved
-
If applicable, the type of the network you're using (like Ethernet 10,
Ethernet 100, Myrinet,...)
-
Any other infos you may find relevant
