Erik Ruf
Erik Ruf
I am a researcher at Microsoft
Research.
Research Interests
My interests lie in the area of design and implementation of programming
languages, especially “higher-level” language features such as type safety,
automated storage management, data/module abstractions, and multithreaded
/distributed programming support. I
believe that such features not only improve programmer productivity, but will
become essential for maintaining performance as machine and network
architectures become ever more complex.
For the past several years, I have concentrated on program analyses
supporting advanced optimizations for object-oriented languages. More recently, my interests have shifted away
from performance issues to the programmer’s interaction with languages,
libraries, and development environments.
Current
Projects
- I am
investigating language design and compilation for data-parallel
applications.
Past Projects
- The Perfect system assisted
programmers in fixing compile-time errors by automatically suggesting
corrections to program source code.
- Improved flow analysis
techniques for context-sensitive modeling of object-polymorphic
containers. A preprint of the SAS’02
paper is available on my publications page.
- Design and implementation of
Bartok, a research abstraction of the .Net programming platform, with
initial work focusing on high level optimizations for C# (see the Advanced Compiler Technology
research group).
- Inter-module static analysis
techniques enabling code and data transformations (improving/eliminating
dispatch, synchronization and storage management operations, object
hierarchy and layout) for Java programs.
A paper on optimizing Java synchronization
appeared in PLDI’2000.
- "Standard"
per-method and inexpensive inter-module optimizations for the Marmot Java compiler.
- The AST Library, which later
became the AST Toolkit, which is now used as the program representation
for a number of internal
software reliability tools.
- Data
specialization: an effective, low-overhead program staging technique.
- Sparse dataflow frameworks
and points-to analyses for the Value
Dependence Graph.
- The FUSE system, a program specializer for
Scheme (Ph.D. thesis, Stanford University,
1993).
Contact information
Address: Microsoft Research, One Microsoft Way, Redmond,
WA 98052
USA
Phone: (425) 706-6111; Fax: (425) 936-7329
Email: my email alias is 'erikruf' at my employer dot com. No spam, please!
Web: http://www.research.microsoft.com/~ruf/
Last modified March 10, 2004
Erik Ruf