Computer science colloquia: Hridesh Rajan
| Date/Time: | Thursday, 26 Aug 2010 at 3:40 pm |
|---|---|
| Location: | B29 Atanasoff |
| Cost: | Free |
| Phone: | 515-294-6516 |
| Channel: | College of Liberal Arts and Sciences |
| Categories: | Lectures |
| Actions: | Download iCal/vCal | Email Reminder |
Software systems must face two challenges today: growing complexity and increasing parallelism in the underlying computational models. The problem of increased complexity is often solved by dividing systems into modules in a way that permits analysis of these modules in isolation. The key challenge in doing so is to manage the explicit and implicit dependence between modules that is known to decrease modularity.
The problem of lack of concurrency is often tackled by dividing system execution into tasks that permits execution of these tasks in isolation. The key challenge in doing so is to manage the explicit and implicit dependence between tasks that decreases parallelism.
Even though these challenges appear to be strikingly similar, current software design practices and languages do not take advantage of this similarity. The net effect is that the modularity and concurrency goals are often tackled mutually exclusively. Making progress towards one goal does not naturally contribute towards the other.
In this talk, I will discuss preliminary efforts of my research group towards achieving synergy between modularity and concurrency goals. Our position is that for programmers that are not formally and rigorously trained in the concurrency discipline the safest and most productive way to get scalability in their software is by improving modularity of their software using programming language features and design practices that reconcile modularity and concurrency goals.