Parallel Processing

Total run-time efficiency (all solution phases) for parallel MLFMM solution of a problem with 3.18 million unknowns.

Many modern computer systems make use of multiple processing units in order to improve computing performance. Such systems include:

• simple multicore CPUs (i.e. one computer with one CPU having multiple cores),
• multi-CPU PCs and SMP workstations (symmetric multiprocessor, typically 2 to 8 CPUs),
• large massively parallel distributed systems with typically 128 to 1024 CPUs (which can again be multicore).

In order to gain the most benefit from the computational hardware, parallel versions of FEKO support state-of-the-art interconnect technologies like GigE, Myrinet, Infiniband or vendor proprietary interconnects like the SGI NumaFlex technology.

In FEKO all the solution phases for all the various numerical techniques have been parallelised, e.g. the ray-tracing for UTD, the MoM matrix setup and solution, the near- and far-field calculations and also seemingly simple things such as power loss computations.

We are very proud of the parallel efficiency of the MLFMM in FEKO. Even for this mathematically complex technique all the phases of the solution process (near-field matrix setup, aggregation, translation, disaggregation, pre-conditioning, iterative solution etc.) have been parallelised rigorously. The efficiency of the parallel implementation in FEKO is in the order of 80% to 95%, depending on the problem and the solution phase etc. This means that for a system with 32 cores the run-time would be approximately 26 times (0.8*32) faster than on a sequential run, i.e. a single core.

Distributed/Shared Memory Hybridisation Optimized

FEKO is designed to use the RAM available in any computing system as optimally as possible.

Distributed memory parallellization breaks up large blocks of memory for simulation on different nodes in a multi-node cluster computing system.  To save processing time FEKO limits communication between processors by making identical copies of information that is often required and sending such a block to each node in the cluster.

Distributed memory parallellization for a multi-core CPU computer

In multi-core CPU and/or server architectures where multiple CPUs and cores are located on the same motherboard, sharing a block of RAM that contains relevant information is the ideal implementation.  Each core can address any memory block in the shared RAM and copies of important information is not necessary to speed up computation.  Such processing architectures are catered for in FEKO with shared memory parallellization.

Shared memory parallellization for a multi-core CPU computer

Distributed and shared memory parallellization is also hybridised in FEKO for multi-node computing clusters made up of multiple muti-core CPUs.

Hybridised memory parallellization in a multi-core two CPU computing system

Intel Cluster Ready

The "Intel Cluster Ready" program facilitates easier design, build and deployment of cluster computers. Intel works with system and software vendors to provide you with certified Linux-based systems clusters based on the Intel Cluster Ready architecture to assure application compatibility and ease of deployment.

FEKO is dedicated to improving the performance of our software in cluster computing environments and work closely with Intel engineers in this endeavour. As such FEKO was first certified as ISV by Intel in 2007 and may since proudly brand our software with the Intel Cluster Ready logo. This means that FEKO customers can purchase an Intel Cluster Ready certified computer with the confidence that FEKO has been qualified on this computing environment and will work straight out of the box.

More information on this initiative can be found on the Intel Cluster Ready website.  These are our preferred Intel Cluster Ready hardware providers:

SGI Cyclone Cloud Computing

SGI high performance computing (HPC) systems have been market leaders in high-end computing hardware, software and networking solutions for over 20 years. The SGI Cyclone Cloud solution was designed to serve the science and engineering communities that rely on very high-end computational hardware.

The SGI Cyclone system offers a pay-per-use usage model for FEKO users that require temporary use of a large computing platform. The system offers large scale parallellization of FEKO simulations in an evironment with sufficient memory for large problems and GPU computation acceleration (with release of FEKO 6.0, 3rd quarter 2010).

Interested customers should contact their local FEKO distributor or SGI via the SGI Cyclone website to discuss options for usage of the Cyclone system.

CrunchYard

CrunchYard offers an engineering software simulation service on a pay per use basis. The service allows scientists and engineers to interactively make use of simulation software via the internet. The benefits of this approach include:

• Reduced cost.
• No infrastructure maintenance is required.
• No hidden costs such as IT support, licensing and computing power.
• Improved performance is shown because simulations are run on a large computing platform.

FEKO users who are interested in testing or using the service should contact their local FEKO representatives or visit the CrunchYard website.

Windows HPC Server Ready

Windows HPC Server provides a productive, cost-effective, and high-performance computing (HPC) solution that runs on x64-bit hardware.  In addition to supporting OpenMP, MPI, and Web Services, Windows HPC Server also supports third-party numerical library providers, performance optimizers, compilers, and a native parallel debugger for developing and troubleshooting parallel programs.

FEKO has been optimised for use on high performance parallel processing systems and as such it is important to FEKO to maintain compatibility with Microsoft's HPC Server.