Accelerating Critical Biomedical Applications Using GPUs
Graphics Processing Units (GPUs) have become the parallel platform of choice in many scientific computing domains, mainly due to their impressive processing capabilities and their attractive cost-performance. Equipped with general purpose programming languages such as NVIDIA's CUDA and Kronos's OpenCL, GPUs have have been able to address computational barriers in a number of critical computational problem.
In this talk we will focus on biomedical applications. Utilizing both AMD and NVIDIA GPUs we have been able to obtain impressive speedups on these cost-effective platforms. This talk will describe some of the applications, and discuss some of our most recent work on GPU compiler and programming tools framework to optimize the execution of some of these key applications.
About the speaker:
David Kaeli is a Full Professor in the ECE Department at Northeastern University. Prior to joining Northeastern, he spent 12 years at IBM. His research looks at the performance and design of high-performance computer systems and software. Current research topics include: profile-guided compilation, GPGPUs, virtualization, intrusion detection, IO performance and reliability modeling, power modeling, database systems, branch prediction studies, workload characterization, memory hierarchy design, embedded systems design, digital signal processors and software testing.