Thursday, February 1, 2001
Noon - 1 pm
Wean Hall 8220
ECE, Carnegie Mellon
Computer Architecture Research in the CMU Impetus Group
In this talk, I will survey ongoing computer architecture research
projects in the Impetus group. First, I will briefly go over three
projects that target: (1) prediction/speculation in high-performance
memory systems, (2) power-aware processor and memory architecture, and
(3) single-chip multiprocessor/multi-threaded architectures.
I will then present details on novel prediction and speculation
techniques to improve memory system performance in shared-memory
multiproceossors. I will present results showing for the first time
that multiprocessors need not support relaxed memory consistency
models to achieve high performance. We identify the key mechanisms
that allow memory to provide the simple and intuitive uniprocessor
programming model while achieving the performance only possible with
the most sophisticated memory models. Next, I will present novel
hardware predictors that capitalize on repetitive memory access
behavior in applications to hide memory latency.
Babak Falsafi joined the Electrical and Computer Engineering
Department at CMU as an Assistant Professor in January 2001.
Prior to joining CMU, he held a position as an
Assistant Professor in the School of Electrical and Computer
Engineering at Purdue University. His research interests include
prediction and speculation in high-performance memory systems,
power-aware processor and memory architectures, single-chip
multiprocessor/multi-threaded architectures, and analytic and
simulation tools for computer system performance evaluation. He has
made numerous contributions in the design of distributed shared-memory
multiprocessors and memory systems, the most prominent of which is
Reactive NUMA, an adaptive memory system that lays the foundation for
the recent prototype by Sun Microsystems code-named WildFire.
He is a recipient of an NSF CAREER award in 2000.
For Further Seminar Info: