David Luebke, Research Scientist, NVIDIA Corporation
IIC Colloquium - Accelerating Science with Massively Parallel Computing
Date: Wednesday, November 12, 2008
Time: 4:00 p.m., refreshments at 3:30 p.m.
Location: 60 Oxford Street, Room 330
Abstract:
Modern GPUs (graphics processing units) provide a level of massively
parallel computation that was once the preserve of supercomputers. NVIDIA's
CUDA platform provides a scalable parallel programming model for GPUs
consisting of minimal but expressive changes to the familiar C/C++ language,
allowing programmers to focus on parallel algorithms rather than the
mechanics of a new programming language. Using this platform, researchers
across science and engineering are accelerating applications by up to two
orders of magnitude. This talk will motivate GPU computing and explore the
transition it represents in massively parallel computing from supercomputing
to commodity "manycore" hardware available to all. In addition to sampling
results to date, the colloquium will address the goals, implications, and
key abstractions of the CUDA programming model and future directions for GPU
computing.
More information:
http://iic.harvard.edu
_______________________________________________
iic-colloquium mailing list
iic-colloquium(a)calists.harvard.edu
http://calists.harvard.edu/mailman/listinfo/iic-colloquium