Meeting co-located with ICSE2010, Cape Town, South Africa
Time and Place
Tuesday, May 4
Time: 6pm (after the Multicore Software Engineering Tutorial T10 @ ICSE)
Location: 2.6(1) (same room as Multicore Software Engineering Tutorial T10 @ ICSE)
Cape Town, South Africa
06.00pm - 06.05pm: Welcome address
06.05pm - 07.05pm: Keynote by Victor Luchangco
07.05pm - 07.15pm: Report on SEPARS
07.15pm - 07.30pm: Discussion
Abstract: With the advent of multicore chips, programmers will need to write concurrent programs to exploit new hardware. But the current state of the practice is too difficult and error-prone for "everyday programming": despite over forty years of experience with concurrent programming, any but the most trivial concurrency and synchronization is typically the domain of expert programmers, and even they often get it wrong. We need new ways to think about concurrent programming, emphasizing data structures and algorithms that are fundamentally parallelizable, and new tools--languages, synchronization constructs, etc.--to exploit the available parallelism, and to reason about and manage the resulting programs. In this talk, I'll discuss some ways in which I think programs and how programmers think about them will need to change, and present some examples using the Fortress programming language, a new language we are developing at Sun Labs.
Bio: Victor Luchangco works in the Scalable Synchronization and Programming Languages Research Groups at Sun Labs, Oracle. His research focuses on developing algorithms and mechanisms to support concurrent programming on shared-memory multiprocessors. He is also involved in the design of the Fortress programming language, a new language aimed at scientific computing. Although a theoretician by disposition and training, Victor believes that systems must address the practical concerns of programmers; he would like to design systems that people will actually use. Because concurrent programming is notoriously difficult and error-prone, he is also interested in exploring how to to simplify reasoning about concurrent systems, both by changing how people design and think about these systems, and by using tools to aid in formal verification. Victor received an ScD in Computer Science from MIT, with a dissertation on models for weakly consistent memories.
See ICSE2010 conference Web site.