===== Abstract ===== **Boosting Large-scale Scientific Simulations with Hybrid Computing Systems based on Graphics Processing Units** **Speakers:** Mark Visconti (NVIDIA) and Joseph Han (Penguin Computing) **Abstract:** In the first part of this talk, Mark Visconti (NVIDIA) will give an overview on Graphics Processing Unit (GPU) architectures and the development of the CUDA programming language with a focus on its impact on High Performance Computing (HPC) applications. This will include an overview of NVIDIA's Tesla product line for HPC and success stories of GPU acceleration for applications in Biology, Oil and Gas, and other scientific research areas. In the second part of this talk, Joseph Han (Penguin Computing) will present three different systems based on multi-core architectures and GPU accelerators. The systems are respectively small, medium, and large budget and power solutions for academic groups. A refreshment will be served at 2:30PM, after the talk, for further discussions with the speakers. **Location, day, and time:** Smith Hall 102A, Friday February 20, 2009, 1:30PM. **Face-2-Face meetings with the speakers:** If you want to better understand GPU potentials for scientific applications, Mark Visconti and Joseph Han are available for F2F meetings on Friday morning. Please contact Michela Taufer at taufer@udel.edu for scheduling a F2F meeting. **Contact:** Michela Taufer (CIS), taufer@udel.edu ===== Speakers' bio ===== **Mark Visconti, Senior Business Development Manager, NVIDIA Corporation** Mark works for NVIDIA's Professional Solutions Group in North America. He provides information and solutions to NVIDIA's government and research customers. Prior to NVIDIA, Mark worked in realtime graphics software development for about six years primarily designing and implementing air and ground visual simulation for the military. He then spent seven years at Silicon Graphics/Cray as a systems engineer helping customers with graphics and HPC solutions. In 2003, Mark joined NVIDIA as an applications engineer. In 2006, he moved into a business development role. Mark has a BSCS from The Ohio State University College of Engineering. **Joseph Han, Ph.D., Director of Systems Engineering and Services, Penguin Computing** Joseph joined Penguin Computing in 2006 after several years at Intel Corporation in the TCAD Integrated Processing Applications Group. As Director of Systems Engineering and Services at Penguin, he is tasked with ensuring that optimal clusters are designed and delivered to exceed customer requirements and expectations. In previous positions, Joseph focused on simulations of kinetics, thermodynamics, and material property prediction and design using ab initio quantum mechanical electronic structure calculations, molecular dynamics and Monte Carlo simulations, and continuum methods. Due to the computationally intensive nature of the simulations, Joseph has been working with Linux and building clusters since 1996. Because of his extensive application specific knowledge of HPC workflows, he has greatly improved customer experiences when running Penguin Computing Clusters with the Scyld ClusterWare management software. Joseph has a B.S. and M.S. from Texas A&M University and a M.S. and Ph.D. from Stanford University in Chemical Engineering.