Harvesting the opportunity of GPU-based acceleration
Matei Ripeanu
Electrical and Computer Engineering Department
University of British Columbia
Stuart Building 212
Wednesday, February 23rd, 2011
11:30AM - 12:30PM
Abstract: Current trends suggest that future high-end systems will be 'hybrid'. Regardless of whether they are high-end workstations or high-performance computing clusters, these systems will include both processors optimized for fast sequential processing and accelerators optimized for massively-parallel processing. One example of this hybrid architecture is a GPU-based platform which integrates commodity processors (CPUs) and Graphics Processing Units (GPUs). These platforms promise intriguing opportunities: compared to traditional computing platforms, they offer a significant increase in peak processing power and memory bandwidth within the same dollar or energy budget, yet they have drastically different architecture and performance tradeoffs. This talk will focus on our experience designing the methodologies and the middleware support to harvest opportunity of GPU-based application acceleration.
Slides (PDF)
Bio: Matei Ripeanu received his PhD in computer science from The University of Chicago in 2005. After a brief visiting period with Argonne National Laboratory, Matei joined the Electrical and Computer Engineering Department of The University of British Columbia. For more information: http://netsyslab.ece.ubc.ca
