----- Original Message ----- From: "Allan Struthers" To: "mathfac-l" , "mathgrad-l" Sent: Friday, March 18, 2011 10:05:29 AM GMT -05:00 US/Canada Eastern Subject: [mathfac-l] Fall 2011: MA5580 - Parallel Numerical Algorithms Hi: Several people have asked me about the topics course MA5580 - Parallel Numerical Algorithms I am running this Fall. Hopefully the following answers some of these questions. Please forward this to anyone who you think might be interested and encourage them to contact me. The course will address design and implementation of parallel algorithms for a number of different fundamental computational areas including Linear Algebra (Sparse and Dense), Numerical ODE., and Numerical Optimization. The course should be accessible to anyone with some computational programming experience. The aspects of computer architecture important to the design of algorithms will be introduced and discussed. When designing and implementing algorithms our primary target will be the very hierarchical design of a modern Graphical Programming Unit GPU with a few hundred hierarchical cores, a hierarchical memory structure including shared memory blocks, a hierarchical thread structure, and a scheduler capable of effectively managing a few thousand threads. Examples will be implemented in either the small CUDA extension to C (run on mathlab machines, the paracuda cluster, or suitable home machines) and/or simulated in Mathematica or Matlab depending on the audience. Please contact me if I can provide any further information. Allan S.