skip to main content
Department of Computer Science University of Colorado Boulder
cu: home | engineering | mycuinfo | about | cu a-z | search cu | contact cu cs: about | calendar | directory | catalog | schedules | mobile | contact cs
home · events · thesis defenses · 2010-2011 · 
 

Thesis Defense - Karlin

 
2/24/2011
12:00pm-2:00pm
ECCS 123

Runtime Prediction of Fused Linear Algebra in a Compiler Framework
Computer Science PhD Candidate

On modern processors, data transfer exceeds floating-point operations as the predominant cost in many linear algebra computations. For these memory-bound calculations, reducing data movement is often the only way to significantly increase their speed. One tuning technique that focuses on reducing memory accesses is loop fusion. However, determining the optimum amount of loop fusion to apply to a routine is difficult as fusion can both positively and negatively impact memory traffic.

In this thesis, we perform an in depth analysis of how loop fusion affects data movement throughout the memory hierarchy. The results of this analysis are used to create a memory model for fused linear algebra calculations. The model predicts data movement throughout the memory hierarchy. Included in its design are runtime and accuracy tradeoffs based on our fusion research. The model's memory traffic predictions are converted to runtime estimates that can be used to compare loop fusion variants on serial and shared memory parallel machines. We integrate our model into a compiler where its predictions often reduce compile times by 99% or more. The kernel produced by the compiler with the model turned on are usually the same as the optimal kernel for the target architecture found by exhaustively testing all possible loop fusion combinations.

Committee: Elizabeth Jessup, Professor (Chair)
Jeremy Siek, Assistant Professor
Xiao-Chuan Cai, Professor
Manish Vachharajani, Assistant Professor
Jonathan Hu, Sandia National Laboratories

 
See also:
Department of Computer Science
College of Engineering and Applied Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
Questions/Comments?
Send email to

Engineering Center Office Tower
ECOT 717
+1-303-492-7514
FAX +1-303-492-2844
XHTML 1.0/CSS2 ©2012 Regents of the University of Colorado
Privacy · Legal · Trademarks
May 5, 2012 (13:40)
 
.