Parallel processing for high-throughput simulation offers fast and economical
computation on clusters of commodity computer hardware. Applications of
parallel computing include solid mechanics, fluid dynamics, ray tracing and
other visualization, molecular dynamics and protein folding, and cellular
automata to model phenomena from epidemiology to options trading. However,
implementation is difficult on many levels, from designing and setting up a
computer cluster to optimizing software for large parallel machines.
Opennovation offers services across this range of challenges
including the following:
Opennovation also owns and operates a cluster with 48 CPUs available as needed
for running simulations.
- Design, setup and maintenance of "diskless" Linux clusters, in which a
single server houses all software including the operating system, and
cluster nodes download everything across the network at boot time or run
time, greatly simplifying cluster administration.
- Selection of software or parallelization tools for a given task.
- Parallelization of existing software to run on multiple processors or
across a cluster, e.g. using MPI (Message Passing Interface).
- Optimizing software to scale efficiently to large numbers of processors.
See also the
distributed visualization library.
- A. Powell, O. Ocholi, J. Vieyra and B. Zhou, "Illuminator: a
Free Graphics Toolkit for Parallel Visualization of Continuum Field Data,"
Poster at IEEE Cluster 2005 (submitted paper).
open innovation nova ovation pen vat ion