Allinea to Enhance DDT Debugging Tool for GPGPU Hybrid and 32K “More-Core” Systems through collaboration with CEA

Agreement with French Atomic Energy Commission (CEA)

anticipate the enhancement of Allinea’s DDT software for HPC platforms.

 

 

Allinea Software has signed a collaboration agreement with CEA, (the French Atomic Energy Commission)to develop enhancements to Allinea’s Distributed Debugging Tool (DDT) for next generation hybrid and “many-core computer systems. Once developed, this technology will be made available to Allinea’s customers.

 

Allinea and CEA will focus their collaboration on two primary projects. The first project is to enhance DDT to be able to debug hybrid GPGPU (general purpose computing on graphics processing units) systems, that transfer most of the intensive computational tasks from standard CPU servers to powerful multi-core, parallel-processing graphics boards to make processing much faster and cost-efficient. The project will begin with NVIDIA’s Tesla GPGPU board, and a fully-functional early demonstration is expected to be shown at ISC09 in HamburgGermany in June, 2009.


“To take full advantage of the speed and computing capacity of these innovative GPU computing systems, and to help wide scale adoption in the market, developers need an easy-to-use tool to debug programs for NVIDIA’s latest line of Tesla computing solutions,” says Andy Keane, general manager of the Tesla business at NVIDIA. “Allinea’s DDT product fills a critical gap for developers and is an ideal tool to support parallel computing with GPUs.”

 

The focus of the second project is to make debugging tools both portable and easy to use for large scale debugging up to 32,000 cores. DDT’s interface is intuitive at every scale of parallelism, and the DDT architecture has already been shown to scale well to existing large systems,” said Dr. David Lecomber, CTO of Allinea Software. As multi-core systems inevitably become ‘many-core systems, the performance and resource consumption of common tools could become significant. It is important to address the growing numbers of cores in clusters and within individual nodes to take advantage of optimizations that are feasible for these systems. This will ensure that developers can use a tool that scales well on large many-core clusters both in terms of performance and in ease of use. We look forward to working with CEA and other HPC sites to develop the right tools for the next generation of computing technology, and to sharing these applications with our customers worldwide.”

 

“CEA has been using Allinea’s DDT and OPT (Optimization and Profiling Tool) products successfully on our large supercomputers, including Bull, Hewlett Packard and NEC systems,” says Hervé LozachComputing Center Operational Manager, CEA. “We chose Allinea’s products because they are an ideal complement to other software development tools. Collaborating with key technology vendors for HPC allows us to enhance and asses the components we need for leading edge computing systems. Another goal is to promote and develop the mastering of HPC technologies in Europe and Allinea’s expertise on parallel performance tools allow us to foster these products, critical for both the large scale and small scale computing market.”

 

Allinea Software will be an exhibitor at the International Supercomputing Conference (ISC2009) in HamburgGermany from June 23-26, 2009.