publications Markus Püschel
Professor
Electrical and Computer Engineering
Carnegie Mellon University
pueschel at ece.cmu.edu
+1 412 268 3804

home

publications

teaching

short CV

personal

the pub

home

Research Areas and Interests

I like to combine techniques from mathematics, computer science, and engineering to attack difficult problems.

My research is divided into two larger longterm efforts that I am leading called SPIRAL and SMART.

SPIRAL is a collection of interdisciplinary projects with the goal to develop new methodologies to automate software (and hardware) development and optimization for specific problem domains (commercial computing, signal processing, scientific computing). In other words, the goal is to "teach"computers to write fast libraries using techniques from mathematics, programming languages, symbolic computation, compilers, and others.

In SMART we are developing a novel approach to linear signal processing, termed Algebraic Signal Processing Theory. We just finished the first papers that start developing this theory. On application is a comprehensive theory of fast transform algorithms.

I am affiliated with CALCM (Computer Architecture Lab), CBI (Center for Bioinformatics), CenSCIR (Center for Sensed Critical Infrastructure Research), CyLab, ITRI (Industrial Technology Research Institute Lab)

Research Group

PhD. students

Project Scientists

Other students I work with:

Latest Teaching

Spring 2009: Introduction to Computer Systems (ECE: 18-243, CS: 15-213)

Collaborators

Pedro Aguiar (IST, Lisbon, Portugal), Basilio Fraguela (U. A Coruna, Spain), Franz Franchetti (CMU), Madeleine Glick (Intel Lab Pittsburgh), James C. Hoe (CMU), Jeremy Johnson (Drexel U., Philadelphia), Robert Killey (UC London), (Jelena Kovacevic (CMU), José M. F. Moura (CMU), Rohit Negi (CMU), David Padua (UIUC), Martin Rötteler (NEC Labs, Princeton), Christoph Ueberhuber (U. Tech., Vienna, Austria), Joao Xavier (IST, Lisbon, Portugal)

Current Projects and Sponsors

Automatic Program Generation for Data-Dependent Applications
ONR
Generating IPP Library Functionality for Larrabee using SPIRAL Intel
FFT Generation for the Cell Processor Mercury
Program Generation for Parallel Platforms NSF, CPA
Algebraic Signal Processing Theory: Towards Multiresolution Analysis NSF, TF
Library Generation for Intel's MKL using Spiral Intel
Intelligent SW/HW Compilers for Signal Processing Applications DARPA
Intelligent HW/SW Compilers for DSP Applications NSF, medium ITR

The Pub

Check out the pub, a tool for maintaining publications and presenting them on the web.

Contact Info

Markus Püschel
Dept. of Electrical and Computer Engineering
Porter Hall B16
Carnegie Mellon University
5000 Forbes Ave
Pittsburgh, PA 15213
phone: +1 412 268 3804
fax: +1 412 268 3890
email: pueschel at ece.cmu.edu
web: http://www.ece.cmu.edu/~pueschel
NSF Discovery
July 2008: Our work on computer generation of high performance libraries (part of Spiral) is selected as NSF discovery
Best Thesis Award
May 2008: My student Yevgen Voronenko wins the ECE/CMU best thesis award. His thesis automates all major aspects of library development for linear transforms.
Algebraic Theory of
Signal Processing
We have developed a new approach to the foundation of signal processing, called Algebraic Theory of Signal Processing.

Fall 2006: The third consecutive NSF grant was just awarded to this effort.
IEEE Proceedings
Special Issue
Have a look at the February 2005 special issue on "Program Generation, Optimization, and Adaptation" of the IEEE Proceedings.

The special issue also contains an extensive SPIRAL overview paper.