[Colloquium] Reminder: Armstrong/MS Presentation/May 24, 2011

[Margaret Jaffey]

A reminder about Tim's MS Presentation tomorrow.

------------------------------------------------------------------------------
Date:  Tuesday, May 24, 2011

Time:  2:00 PM

Place:  Ryerson 276

M.S. Candidate:  Tim Armstrong

M.S. Paper Title: Integrating Task Parallelism into the Python
Programming Language

Abstract:
One of the major current challenges in computer science is providing
programming models and abstractions that allow efficient and correct
parallel programs to be expressed. The challenge occurs in a number of
domains, with different scales of paralellism required. In the
consumer space, eight or ten cores will soon be common on desktop
computers, and the performance of software which cannot take full
advantage of many cores will lag. In the scientific space, scientific
investigations increasingly require computationally intensive
simulation and data crunching on computer clusters.

Coordination languages are one way in which the problem of writing
parallel programs can be made more manageable. A coordination language
provides a specialized way to specify the communication and
synchronization between the different components of a parallel or
distributed application. In many cases, the parallelism in a program
is naturally expressed as task parallelism, where independent tasks
which produce and consume data can run in parallel.

This paper proposes and demonstrates the feasibility of embedding a
task-parallel coordination language, PyDFlow, into the general-purpose
scripting language Python. PyDFlow currently supports loosely coupled
workflows, where tasks communicate through input and output files, but
the language is designed to be extensible beyond this.

PyDFlow represents parallel computations as graphs of tasks linked by
data dependencies and provides a programming model somewhat akin to
functional languages, where computations are specified by constructing
expressions by composing functions. This gives natural syntax and
semantics for expressing a parallel or distributed computation that
integrates well with the Python language, and allows execution
techniques from the functional language literature to be applied.

Tim's advisor is Prof. Ian Foster

Login to the Computer Science Department website for details:
 https://www.cs.uchicago.edu/phd/ms_announcements#tga

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Margaret P. Jaffey            margaret at cs.uchicago.edu
Department of Computer Science
Student Support Rep (Ry 156)               (773) 702-6011
The University of Chicago      http://www.cs.uchicago.edu
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=