!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
                                       
                   B O S T O N   U N I V E R S I T Y
                                       
                      Computer Science Department
                                       
                          C O L L O Q U I U M

           An Optimizing Compiler for Relaxed Memory Consistency Models
    
                            Jaejin Lee

              Computer Science and Engineering Department
                      Michigan State University

                            Wednesday, March 14
			       11:00 am
                       (Coffee served at 10:45PM)

                         Seminar Room / MCS 135

While memory consistency models has long been an interesting topic in
the computer architecture community, the usability of memory
consistency models, their impact on performance, and the compiler
technology needed to optimize parallel programs are poorly understood
in the programming languages and compilers community. A concrete
example is the current Java memory model that is one of relaxed memory
consistency models. The Java memory model is very difficult for
programmers to understand, and there are several ways of interpreting
it.

This talk focuses on building an optimizing compiler for explicitly
parallel shared memory programs that hides the underlying relaxed
memory consistency model. The compiler presents an intuitive and
natural memory consistency model (e.g., sequential consistency) to the
programmer. It shifts the programmer's burden of considering the
underlying machine architecture to the compiler itself so that it
makes programming and debugging easier. Moreover, it provides correct
compiler level optimizations that cannot be correctly applied to
parallel programs by conventional compilers. The compiler uses Shasha
and Snir's delay set analysis to hide the underlying relaxed memory
consistency model. To extend classical compiler optimization
techniques to parallel programs, it uses the Concurrent Static Single
Assignment form and Concurrent Global Value Numbering.

The compiler will serve as a testbed to prototype new memory
consistency models at the language level, and to measure the effects
of different memory models and different compiler optimizations on
program performance.


Bio:
Jaejin Lee received his Ph.D. degree in Computer Science from the
University of Illinois at Urbana-Champaign in 1999. He was a
recipient of an IBM Cooperative Fellowship and a scholarship from the
Korea Foundation for Advanced Studies during his Ph.D. study. He
received an M.S. degree in Computer Science from Stanford University
in 1995 and a B.S. degree in Physics from Seoul National University
in 1991. He spent six months at the University of Illinois
at Urbana-Champaign as a visiting lecturer and postdoctoral research
associate. Currently, He is an assistant professor in the Computer
Science and Engineering department at Michigan State University.

Host: Richard West
-------------------------------------------------------------------------------
For colloquium info, including directions, see http://cs-www.bu.edu/colloquium
-------------------------------------------------------------------------------