Login
Computer Science @ Rochester
Monday, April 23, 2012
11:30 AM
Computer Studies Bldg. Room 209
Ayon Basumallik
MathWorks Inc.
Code Generation and Runtime Support for Dynamic Languages - Challenges and Opportunities
Dynamic Languages have steadily grown in popularity over the past decade, especially in the field of computational sciences. Owing to more intuitive syntax and comprehensive functionality provided by included libraries, dynamic languages such as MATLAB(r) are regularly used by over a million users. The same features that contribute to ease-of-use and programmer productivity also present novel challenges and opportunities for innovation in the design and implementation of compilers and runtime support.

This presentation will explore the landscape of dynamic languages, unique features that characterize dynamic languages and the challenges involved in supporting these features. We shall take an in-depth look at the ramifications of dynamic typing and discuss some of the proposed approaches to code generation for dynamically typed languages. We shall then explore some other features of dynamic languages such as introspection and runtime evaluation that constrain the applicability of traditional compiler optimizations. Finally, we shall examine the challenges in providing runtime support for features such as automatic memory management.

Dr. Ayon Basumallik is a Senior Engineer and Team Lead at MathWorks Inc. Over the past decade, the goal of his research has been to improve programmer productivity and provide easier programming interfaces for a variety of hardware platforms via innovations in the design and implementation of programming languages. He received his PhD in Computer Engineering from Purdue University. His doctoral dissertation explored ways of making compute clusters easier to program by extending the ease-of-use of OpenMP shared-memory programming to distributed-memory systems. He is currently based in Natick, MA as a member of the core MATLAB Language Team at MathWorks Inc. MATLAB(r) is used by over a million users worldwide as the de facto language for technical computing, in diverse domains such as science, engineering, finance and manufacturing. The MATLAB Language Team is responsible for the evolution and implementation of the MATLAB(r) language for efficient use of current and emerging computing platforms.