Parallel Programming Services logo

Welcome to Parallel Programming Services

Do you want to take full advantage of the power of new multi-core processors?

Do your applications currently only run on sequential or shared-memory parallel systems?

Would you like your numerical models to solve larger problems in faster times - larger and faster by orders of magnitude?

If so, we can help you do this!

All modern computers now contain multi-core processors, with 4 or more physical compute-cores per chip. Meanwhile, clock-speeds are not increasing past 3 GHz or so. The only way to speed up a single job is to parallelize it, by "multi-threading" it at the very least.

We can help you transform your applications by programming them either for multi-core processors or for more general distributed-memory parallel systems.

Parallel programming based on multi-threading allows your sequential applications exploit the full power of multi-core processors and multi-processor shared-memory systems.

Parallel programming based on domain decomposition (a "divide and conquer" approach), allows applications to avail of the order-of-magnitude speedups that are offered by both clusters and shared-memory parallel systems. Once parallelized, your applications will run on commodity-based industry-standard clusters, or any other distributed-memory parallel system, all of which are becoming ever more powerful, and more accessible.

distributed-memory cluster
A typical distributed-memory cluster.
One or more racks can contain hundreds of
processors, along with accompanying memory,
storage, and interconnecting fabric
Our parallel programming is done with a very light touch - your code will remain easily recognizable, and you can continue to develop it as before.

We are a group of scientists with backgrounds in physics, chemistry and meteorology, each of us with proven expertise in parallel programming and many years experience in the high-performance computing industry.

We are ready to collaborate with you to parallelize your flagship applications and provide other high-performance computing consultancy services. Our services range from consulting and training, up to an end-to-end parallelization of your applications to run efficiently on general distributed-memory systems.

There is no need to limp along any more using your old sequential numerical models. With our help, you can turbo-charge them by converting them to run in parallel.

For further information, use the links below to browse the rest of this site. Contact us directly at , or call Enda O'Brien in Ireland at +353 87 751 7969.