High-performance computing is vitally important to the mission of BAE Systems, one of the largest defense, aerospace, and security contractors in the world. As a high-tech manufacturer, BAE Systems relies on simulation and modeling software to design and build its products. The group responsible for identifying and developing this technology inside BAE Systems is called the Advanced Technology Centre.
“We are at the cutting edge of High Performance Computing and use it in areas such as image processing, real time data analysis and simulation whilst using very large data sets to support the design of air, land and sea based platforms,” states Pierre Moinier, group leader for Integrated Modelling Technologies at BAE Systems Advanced Technology Centre.
ATC’s software development team is tasked with delivering quality, reliable software on time to help the UK-based multinational stay competitive against rivals, including defense giants Lockheed Martin and Boeing, and other contractors, such as Raytheon, General Dynamics, and Northrop Grumman.
As a productivity-boosting measure, the software team turned to the Allinea MAP scalable performance profiler to help with the optimization of parallel code running on x86 processors and hybrid CPU-accelerator setups. The ATC had already been using the debugging tool, Allinea DDT, for a number of years and MAP shares the same easy-to-use interface and look and feel, so the developers can be comfortable using it from the start.
Says Moinier: “Our developers are really good at what they do but the speed and throughput of simulations is becoming more important, meaning we need a good profiler that can work with our production and research codes and help us to optimize that performance.”
BAE Systems has long been a leading industrial user of HPC, and is planning to further expand its supercomputing capabilities to facilitate ever more accurate and detailed simulations. Moinier reports that Allinea’s software is “very easy to use even on complex problems and we know this will become even more important as our usage of high performance computing increases.”
Code Acceleration
The ATC has a lot of experience accelerating modeling applications and was an early adopter of GPU technology back in 2008. Moinier reports that his staff developed one of the first computational fluid dynamics codes in the world capable of harnessing the full power of multicore processors and accelerator boards. GPUs provided the engineers with a 10X speedup on that homegrown CFD software.
Now the group is experimenting with the current generation of Xeon Phi coprocessors and thinking about the future generation Phi chips as well. In a recent interview with EnterpriseTech‘s Timothy Prickett Morgan, Moinier reports it’s too early to tell whether the Tesla GPU or Xeon Phi approach will prevail at BAE Systems.
“We just ported our CFD code to the Xeon Phi, and it wasn’t a very difficult activity,” he explains. “However, we have not achieved the top performance yet simply because we have not had the time to tune. In order to achieve the performance you need to spend a lot of time in order to make the best use of vectorization. The code has been ported but it needs to be optimized. The Tesla GPU and Xeon Phi both have clear benefits, but they both come at a price. Our code cannot run directly on GPUs, and with Xeon Phi it might be a little bit easier, but you still have to spend a lot of time in terms of optimization. So at the moment, we haven’t made up our mind and we are not sure who is going to dominate the HPC market. But we should be pretty much ready for whatever will be the most appropriate at the time. The important thing is that both have performance gains and we want to have the choice. So we have invested so we have the right tools and the right expertise.”