Computer Engineering Seminar

Transactional Coherence and Consistency (TCC)

Kunle Olukotun

The key issue facing future parallel systems is the difficulty of writing parallel programs to run on them. Transactional Coherence and Consistency (TCC) is a new way to implement cache coherency in shared-memory parallel systems by using programmer defined transactions as the fundamental unit of parallel work, communication, coherence, consistency and failure atomicity. TCC has the potential to simplify parallel programming by providing a smooth transition from sequential programs to parallel programs. In this talk, I will describe TCC and explain how to develop parallel applications using the TCC programming model. I will also present the results of a performance comparison between TCC and conventional snoopy cache coherence on a suite of well-tuned parallel applications. These results demonstrate that TCC can provide
programming advantages without compromising parallel performance.

Kunle Olukotun is an Associate Professor of Electrical Engineering and Computer Science at Stanford University. Olukotun received his Ph.D. in Computer Science and Engineering in 1991 from The University of Michigan. Olukotun led the Stanford Hydra single-chip multiprocessor research project. Hydra implements a novel architecture for combining multiple processors on a single chip. One of the most innovative aspects of
the Hydra architecture is the support for thread-level speculation. Olukotun founded Afara Websystems to develop commercial server systems with chip multiprocessor technology. Afara was acquired by Sun Microsystems; the Afara processor technology, called Niagara, will appear in future Sun products. Olukotun currently leads projects in
computer architecture, parallel programming environments and formal hardware verification.

Sponsored by