Programming the Finite Element Method

The 5th edition of the text book "Programming the Finite Element Method" is now available. It is aimed at undergraduates, graduate students and practitioners in all areas that use the finite element method. If you have used earlier editions for teaching, now is the time to update reading lists and to get your library to reorder.

Read more ...

Image-based engineering on HPC systems made easy

Press release 29 April 2014. Simpleware and ParaFEM have agreed to collaborate on making it easier to carry out image-based modeling on HPC systems. Currently, there is no common model format shared by the two packages, meaning users have to rely on running conversion scripts at the Linux command line and manually editing files. This can be both time consuming and prone to errors. Therefore both Simpleware and ParaFEM will implement new import/export options for a common data format. Importing and exporting large files can be a bottleneck in parallel processing, so the chosen interface will likely be through the binary Ensight Gold or ExodusII format. These are supported by the open source ParaView visualization package. The collaboration will provide a seamless meshing, solution and postprocessing pipeline for simulations requiring the use of HPC facilities. The common data format capability will be released towards the end of the year.

About Simpleware

Simpleware develops industry-leading software solutions for the conversion of 3D images (as obtained from MRI, CT, micro-CT...) into high-quality Finite Element (FE), Computational Fluid Dynamics (CFD), CAD and 3D Printing models. Simpleware software is being used by engineers and researchers in a variety of fields, including: Life Sciences, Materials Science, Oil and Gas Exploration, Industrial Non-Destructive Evaluation and Reverse Engineering. Simpleware software is based on proprietary technology that facilitates the integration of image data with CAD and computational analysis (FEA and CFD) workflows. Unlike other approaches, Simpleware provides a direct route between 3D image data and simulation-ready models. For more information on Simpleware, please visit:

About ParaFEM

ParaFEM is an open source library for parallel finite element analysis. It is an extension of the software provided in Smith IM, Griffiths DV and Margetts L “Programming the Finite Element Method”, 5th Edition, Wiley, 2014. ParaFEM uses an element-by-element solution strategy together with iterative solvers to solve problems with up to 1 billion degrees of freedom and has been used on systems with up to 64,000 cores. It is under continuing development supported by both industrial and research funding. For more information on ParaFEM, please visit:

Read more ...

One billion equations

Sunday 20th January 2013 is a historic date for ParaFEM. During the middle of the night, we managed to solve a problem with more than 1 billion equations, 1,023,468,720 equations to be precise!

The problem was a trivial one, a linear elastic analysis comprising a grid of 440 by 440 by 440 twenty noded hexahedral elements. However, the program did not take advantage of the regular geometry and therefore should be able to solve an arbitrarily shaped domain that is just as large.

We used HECToR to run the analysis. On 4096 MPI processes, the problem solved in less than 15 minutes. Using 8192 MPI processes, run time was 7 minutes. The compute nodes needed to be run in an "under-populated" state in order to access sufficient memory. One core on each of 4096 16-core processors was used for the 4096 process run. Two cores per processor were used in the 8192 process run.

Later in the year, we hope to report results for an arbitrary geometry and visualize the results!

Seminar: Hackers, Supercomputers and Faster Finite Element Modelling

Dr Lee Margetts will be giving a talk on ParaFEM, with question and answer session, on Wednesday 12 December, 2-3pm, Room C53, Sackville Street Building, University of Manchester.

Are you using a commercial finite element analysis package? Do your finite element analyses take a very long time to run? Have you simplified your model so much that it no longer represents the physics of your problem? Are these issues serious limitations to your research output? Supercomputers now have hundreds of thousands of processing elements and provide programs with access to terabytes of memory. This means it is theoretically possible to solve very large and/or complex problems in a reasonable time. Unfortunately, the gap between commercial finite element software and the capability of supercomputing facilities is widening. This talk will discuss whether an army of “hackers” could fill that gap by developing open source parallel finite element software. In particular, the speaker will introduce the ParaFEM project and an accompanying textbook to be published in 2013.

Cloud Computing at NAFEMS Awareness Seminar

Dr Lee Margetts recently gave a presentation on Cloud Computing at the NAFEMS Awareness Seminar on "Future Directions for Engineering Simulation" in Madrid. In his talk, Dr Margetts gives an introductory overview of Cloud Computing and then explains how a Cloud-based front end for ParaFEM will be developed in collaboration with the European funded Venus-C consortium.

Julich Supercomputing Center Boots Up GPU Cluster

Germany's Julich Supercomputing Center (JSC) has gone live with its new Julich Dedicated GPU Environment (JUDGE) cluster, which will be used for ensemble simulations in climate and atmospheric research, as well as for data analysis and simulations on big data sequences in biology and brain research.

JUDGE will enable JSC to optimize the applications for the highest performance. The hybrid system uses graphical general processing units (GPGPUs) and conventional processors. GPGPUs can help boost performance without significantly increasing energy consumption, which is important because improvements in energy efficiency will allow further supercomputing advancements in the future. NVIDIA's Stefan Kraemer says the JUDGE cluster is a good example of how computers need to continue to be developed in the future, following the target of exascale computing. "This is valid not only in regard to performance, but also to energy consumption and energy efficiency," he says. "Pilot projects like JUDGE play a key role in this process and are a key step on the way to hybrid systems.

Source HPCwire: May 07, 2011

Follow the link for the full article

Oak Ridge Supercomputers Modeling Nuclear Future

During the annual televised “State of the Union” address at the beginning of 2011, Barak Obama sought to renew the national focus on science and technology, in part by using supercomputing capabilities to drive progress.

To highlight the role of HPC in the new generation of scientific endeavours, the President told millions of Americans about how supercomputing capabilities at Oak Ridge National Laboratory (ORNL) will lend the muscle for a Department of Energy initiative “to get a lot more power out of our nuclear facilities” via the Consortium for Advanced Simulation of Light Water Reactors.

For further details see the full article here.

Free Trial of Amazon Web Services

High Performance Computing in the Cloud - Amazon are offering a free trial of their Web Services.

This technology might be of interest to ParaFEM users or developers who do not have access to their own HPC systems. In essence, Amazon Web Services allow you to build a virtual cluster on the web and operate it on a pay-as-you-go basis.

For further details and to see a video tutorial explaining how to set up an HPC environment in less than 10 minutes, click here.

Special Issue on Data Intensive Computing in the Clouds

Springer Journal of Grid Computing

Applications and experiments in all areas of science are becoming increasingly complex and more demanding in terms of their computational and data requirements. Some applications generate data volumes reaching hundreds of terabytes and even petabytes. As scientific applications become more data intensive, the management of data resources and dataflow between the storage and compute resources is becoming the main bottleneck.

Analyzing, visualizing, and disseminating these large data sets has become a major challenge and data intensive computing is now considered as the "fourth paradigm" in scientific discovery after empirical, theoretical, and computational scientific approaches.

Read more ...

Google Exacycle for Visiting Faculty Grant Program

Google Exacycle for Visiting Faculty is a new grant program for high-performance, CPU-intensive computing. In its first year, the program invites proposals for large-scale, computationally intensive research projects. The program awards sizable allocations on Google’s computing infrastructure to address grand challenges in science and engineering. We will award a total of approximately one billion core-hours to drive transformational research in diverse fields such as astronomy, biology and medicine, earth sciences, mathematics and physics.

For further details, see

GPU support

Margetts and Bane, IT Services for Research, Univ. of Manchester, have been awarded time on the HECToR GPGPU testbed for developing efficient multi-GPU solvers for use with ParaFEM between mid-April and mid-October 2011.

Updates will follow...

Welcome to ParaFEM

ParaFEM is a freely available, portable library of subroutines for parallel finite element analysis.

The subroutines are written in FORTRAN90/95 and use MPI for message passing. It is an extension of the software developed in Smith I.M. and Griffiths D.V. "Programming the Finite Element Method", Wiley, 2004.