HPCwire: Aalborg University in Denmark Taps Jacket

Since 1986 - Covering the Fastest Computers
in the World and the People Who Run Them

Translation Disclaimer

Subscribe | Sign In

Visit additional Tabor Communication Publications

Leading HPC
Solution Providers

February 15, 2010

Aalborg University in Denmark Taps Jacket

ATLANTA, Feb. 15 -- Researchers at Aalborg University (AAU) in Denmark are applying new GPU technology to the communications field in a quest to advance theory, modeling, design, implementation, and test of RF systems and circuits.

Aalborg's Department of Electronics Systems and Technology Platforms Section focuses on the theory, modeling, design, implementation and testing of RF systems and circuits, mainly within RF-CMOS, HW/SW co-design and design space exploration, advanced circuit and system theory, modeling and analysis as well as synthesis. Recently, a significant effort has been started on the use of GPUs to solve numerical problems in the communication technology area, where the ultimate goal is to improve communication systems and devices.

As a stepping stone towards broader use of GPU technology, Professor Torben Larsen's team at AAU has employed a number of systems housing various GPU configurations to explore the possibilities. The systems include:

Intel Core i7 975 (3.33GHz) with nVidia Quadro FX-3800 + nVidia GeForce 9800HT + nVidia C1060 Tesla GPUs running Windows 7
Apple MacBook Pro with Intel Core 2 Duo (2.8 GHz) and nVidia GeForce running Ubuntu Linux
Apple MacBook Pro with Intel Core 2 Duo (2.8 GHz) and nVidia GeForce 9400M and 9600M GT running Snow Leopard
Colfax CXT-3000 with four nVidia C1060 Tesla running Windows Vista
Colfax CXT-5000 with three nVidia C1060 Tesla and one nVidia Quadro FX-5800 running Windows Vista
Colfax GPU Cluster with 18 Intel X5570's and 16-24 nVidia C1060 Teslas running 64 bit Red Hat Linux, which will be deployed in February this year -- with a 20 Gb/s InfiniBand backbone connecting all nodes

The researchers' primary interest is the development of improved algorithms and techniques that lead to useful discovery, which is why they have chosen to develop in a very high-level language (VHLL). The algorithms will be developed on these systems using Jacket and MATLAB. The development of new algorithms that incorporate Jacket from the beginning means that there is not a "transformation" process to get GPU results. Upon completion of the development, users can either run plain MATLAB on CPU-based systems or leverage the above GPU resources to instantly gain performance advantages.

Theory, modeling, design, implementation and testing of RF systems and circuits are essentially to address advanced signal processing applications. Signal processing is a natural application area for GPUs and well suited within the Jacket environment. The types of computations performed in signal processing include Power Series, Singular Value Decomposition (SVD), Fast Fourier Transforms (FFT), Basic Linear Algebra Subprograms (BLAS), and optimization among others.

The signal processing community is constantly looking for better ways to improve signal processing run times to reach real-time results necessary to deliver on production and research -- missions and objectives. The combination of the latest GPU technology, Jacket, and the familiar MATLAB environment has allowed AAU to explore more cost effective and productive ways to execute on research objectives and in turn deliver results to their funding organizations, such as the Danish Research Council for Technology and Production, The Danish Research Ministry, The European Union's Framework Programs, and the Danish National AdvancedTechnology Foundation, in addition to companies like Infineon Technologies, Agilent, Texas Instruments, ETI, RF Micro Devices, and Neurodan.

Professor Larsen's Section has recently received two large research projects in open competition and will before second quarter 2010 have 10 researchers working on different aspects of GPU computing for communication system modeling and simulation. In all cases it will be based on the Jacket computing platform. "We often have few production runs and frequent model changes, which means that Jacket is an ideal solution for us," says Professor Larsen. He continues saying, "With Jacket, we gain a lot of performance for a very modest investment in learning to take advantage of Jacket. We can't afford to use a lot of time to develop C/FORTRAN programs -- we rather let the computers together with Jacket do the hard computations. The speed-ups we have seen from Jacket compared to CPU solutions are very, very impressive."

Jacket delivers GPU computing power to domain professionals that have little to no CUDA programming background and would prefer to delay or eliminate parallel programming or low level programming languages like C and C++. It enables faster prototyping and problem solving across a range of signal processing -- as well as financial, scientific, and engineering -- applications. Jacket eliminates the need to re-program applications into complex languages which would otherwise require advanced programming knowledge and months to complete.

The adoption of Jacket by Larsen's team at AAU has resulted in a unique collaboration between AccelerEyes and AAU. Dr. Larsen provides continuous feedback to the engineering and development team at AccelerEyes to advance Jacket. Most recently Dr. Larsen has launched Torben's Corner on the AccelerEyes Wiki to provide valuable information to the Jacket user community, globally. Torben's Corner provides Tips and Tricks, a Performance Index, and more, for users to gain relevant information directly from another user of Jacket. AccelerEyes and AAU are very excited about providing this resource to Jacket users.

"I hope that Torbens Corner will produce material, which is of interest to the Jacket community. From the beginning I plan on making material for two areas: 1) Tips and Tricks and 2) Jacket Performance Index. The former is as the name says, a plan to show different examples of what Jacket can be used for. The latter will present a rating for how suitable different functions are to be computed by the GPU. In all cases the focus will be on practical uses of Jacket where all source code is public available," says Larsen. "Having a strong collaboration with Dr. Larsen and his team has already made a big difference for other Jacket users and provides validation for Jacket," said David Gibson, AccelerEyes' vice president of sales and marketing. "AAU is proving to be a leader in the adoption of rapid application development for GPU computing. Our software will help them extend that lead."

About AccelerEyes

AccelerEyes launched in 2007 to commercialize Jacket, the first software platform to deliver productivity in GPU computing. With advanced language processing and runtime technology to transform CPU applications to high performance GPU codes, Jacket extends from desktop workstation performance to also fully leverage GPU clusters. Based in Atlanta, GA., the privately held company markets Jacket for a range of defense, intelligence, biomedical, financial, research, and academic applications. Additional information is available at www.accelereyes.com.

-----

Source: AccelerEyes LLC

Share Options

Subscribe

Subscribe to HPCwire

Discussion

There are 0 discussion items posted.

Join the Discussion

Become a Registered User Today!

Registered Users Log in join the Discussion

RSS Feeds

Feeds by Topic

Feeds by Industry

Feeds by Content Type

Subscribe to All Content

All

May 23, 2012

May 22, 2012

May 21, 2012

May 18, 2012

May 17, 2012

May 16, 2012

May 15, 2012

May 14, 2012

Feature Articles

NVIDIA Works On CPU Co-Dependency Issues with Kepler GPU

NVIDIA is telling everyone that the GK110, its new Kepler GPU aimed at supercomputing, is all about improving performance per watt. But the other driving theme behind the new architecture is reducing the GPU's reliance on its CPU host. How well it accomplishes both these goals areas could determine the success of the new chip in high performance computing.
Read more...

OpenACC Starts to Gather Developer Mindshare

PGI, Cray, and CAPS enterprise are moving quickly to get their new OpenACC-supported compilers into the hands of GPGPU developers. At NVIDIA's GPU Technology Conference this week, there was plenty of discussion around the new HPC accelerator framework, and all three OpenACC compiler makers, as well as NVIDIA, were talking up the technology.
Read more...

NVIDIA Launches Kepler Into HPC

NVIDIA has introduced its first Kepler-generation GPU product for high performance computing, and revealed some of the inner working of the new architecture. The announcement took place at the kickoff of the company's GPU Technology Conference taking place this week in San Jose, California.
Read more...