UTCS Colloquia - Franz Franchetti, Professor, Carnegie Mellon University, "Spiral:Automatic Generation of Industry Strength Performance Libraries"

Contact Name: 
Robert A. Van De Geijin
Location: 
ACES 6.304
Date: 
Jan 25, 2013 1:00pm - 2:00pm

 

Talk Audience: UTCS Faculty, Grads, Undergrads, Other Interested Parties

Host:  Robert van de Geijn and Don Batory

Talk Abstract: The SPIRAL system (www.spiral.net, www.spiralgen.com) is a software production system that automatically generates highly efficient software for important kernel functionality on modern processor architectures. It targets machines across the spectrum from embedded and mobile devices through desktop and server-class machines up to supercomputers. For signal and image processing and communication functions such as the fast Fourier transform (FFT), convolution/correlation, or a Viterbi decoder, SPIRAL has proven to automatically generate code that is tuned to the given processor and computing platform, and is outperforming expertly hand-tuned implementations. Performance metrics include computational efficiency and execution rate, energy efficiency, code size, or a composite of the above.

In this talk, we give a short overview on the core SPIRAL technology.  We then explain how Spiral generates efficient programs from embedded and mobile platforms to supercomputers, targeting homogeneous and heterogeneous multicore processors, field-programmable gate arrays, fixed-function units, and graphics processors (GPUs).  We discuss how issues of code quality, verification, rapid prototyping and high configurability with respect to functionality and interfaces is handled. We conclude with a representative set of performance results across a relevant space of kernel functionality and platforms.

Speaker Bio: Franz Franchetti is Associate Research Professor with the Department of Electrical and Computer Engineering at Carnegie Mellon University and CTO of SpiralGen, a CMU spin-off company commercializing the technology developed in the SPIRAL project. In 2006 he was member of the team winning the Gordon Bell Prize and in 2010 he was member of the team winning the HPC Challenge Class II Award.  Dr. Franchetti's research focuses on automatic performance tuning and program generation for emerging parallel platforms, including embedded processors and mobile chipsets, multicore and manycore CPUs, clusters and high-performance systems, graphics processors, and field programmable gate arrays. His research goal is to enable automatic generation of highly optimized software for important kernel functionality across a wide range of applications and platforms, making a step towards solving the porting problem.

 

Tags: