NB: *** THE ROOM HAS CHANGED TO LECTURE THEATRE LC108*** UNIVERSITY OF HERTFORDSHIRE COMPUTER SCIENCE RESEARCH COLLOQUIUM "A Practical Demonstration of DIMES, a Cellular Architecture" speakers: Jason McGuiness and Colin Egan (Computer Science, University of Hertfordshire) 21 January 2004 (Wednesday) Lecture Theatre LC108 College Lane Campus, Hatfield 3 - 4 pm Coffee/tea and biscuits will be available. Everyone is Welcome to Attend Abstract This talk will continue last week's introduction to Cellular Architectures and DIMES with applications and algorithms. Depending on the possibility of network access, a practical demonstration via access to DIMES at the University of Delaware will be shown. However, we may be able to discuss in more detail the algorithms and provide a simulation. There is an ever-widening gap between CPU speed and memory speed, resulting in a 'memory wall' where the time for memory accesses dominate performance. Cellular architectures, such as the Cyclops family, have been developed to overcome this 'memory wall' by implementing processors-in-memory (PIM) on the same chip. PIM architectures achieve high performance by increasing the bandwidth of processor-memory communication and reducing latency. DIMES (the Delaware Iterative Multiprocessor Emulation System) is being developed, by CAPSL at the University of Delaware, as a hardware validation tool for cellular architectures. The version of DIMES we have used is a simplified hardware implementation of the Cyclops-64 cellular architecture that has been developed at the IBM T. J. Watson Research Center. Since DIMES is a hardware validation tool, its hardware implementation is constrained to a dual processor where each processor has four thread units. DIMES memory is restricted to 16K of local scratch-pad memory per processor and 128K global shared memory. Additionally DIMES is linked to a host computer for I/O. We have chosen to use a Mandelbrot-set generator (written in C++) with a work-stealing algorithm as our metric to evaluate the programming model on DIMES. The Mandelbrot-set generator has been threaded, and the work-stealing algorithm achieves load balancing between the DIMES threads. Our Mandelbrot example demonstrates the effective use of DIMES threads, the effective use of DIMES scratch-pad memory and the effective use DIMES global memory in its RTS environment. The results of the study are highly promising and show that DIMES is an ideal hardware tool for validating future Cyclops enhancements. This is joint work of Jason McGuiness(1,2), Colin Egan(2), and Guang Gao(1). 1 University of Delaware, Newark, DE. USA 2 University of Hertfordshire, UK ------------------------------------------------------------------ Computer Science Colloquium Abstracts On-line: http://homepages.feis.herts.ac.uk/~nehaniv/colloq/