|
Parallel Inclusion-based Points-to Analysis
|
|
A points-to analysis is a static analysis technique that identifies which variables might be pointed by the pointers of the input program. Because this compiler phase can be time consuming, there has been much research work focused on making it faster. We are the first ones exploring a complementary approach: parallelization. This is not only the first implementation of a parallel points-to analysis, but also a fast implementation: in a 16-core machine, we achieve an average speedup of 6x with respect to a state-of-the-art sequential implementation. Furthermore, we also implemented an alternative GPU version of the same algorithm that results on an average speedup of 7x. You can access the GPU (CUDA) code here, and the CPU (Java) code here.
|
|
|
|
|
|
PPoPP'12
|
A GPU Implementation of a Inclusion-based Points-to Analysis
M. Méndez-Lojo, M. Burtscher, K. Pingali
[pdf]
[BibTeX]
[slides]
|
|
PLDI'11
|
The Tao of Parallelism in Algorithms
K. Pingali, D. Nguyen, M. Kulkarni, M. Burtscher, M. Hassaan, R. Kaleem, T. Lee, A. Lenharth, R. Manevich, M. Méndez-Lojo, D. Prountzos, X. Sui
[pdf]
[BibTeX]
[slides]
|
|
OOPSLA'10
|
Parallel Inclusion-based Points-to Analysis
M. Méndez-Lojo, A. Mathew, K. Pingali
[pdf]
[BibTeX]
[slides]
|
PPoPP'10
(best paper nominee)
|
Structure-driven Optimizations for Amorphous Data-Parallel Programs
M. Méndez-Lojo, D. Nguyen, M. Kulkarni, M. Burtscher, M. Hassaan, D. Prountzos, X. Sui, K. Pingali
[pdf]
[BibTeX]
[slides]
|
|
BYTECODE'09
|
User-Definable Resource Usage Bounds Analysis for Java Bytecode
J. Navas, M. Méndez-Lojo, M. Hermenegildo
[pdf]
[BibTeX]
|
|
PASTE'08
|
Sharing Analysis of Arrays, Collections, and Recursive Structures
M. Marron, M. Méndez-Lojo, M. Hermenegildo, D. Stefanovic, D. Kapur
[pdf]
[BibTeX]
[slides]
|
|
LCPC'08
|
Efficient Set Sharing using ZBDDs
M. Méndez-Lojo, O. Lhoták, M. Hermenegildo
[pdf]
[BibTeX]
[slides]
|
|
VMCAI'08
|
Precise Set Sharing Analysis for Java-style Programs
M. Méndez-Lojo, M. Hermenegildo
[pdf]
[BibTeX]
[slides]
|
|
LOPSTR'07
|
A Flexible, (C)LP-based Approach to the Analysis of Object-Oriented Programs
M. Méndez-Lojo, J. Navas, M. Hermenegildo
[pdf]
[BibTeX]
|
|
FTfJP'07
|
Generic, Context Sensitive Analysis Framework for Object Oriented Programs
J. Navas, M. Méndez-Lojo, M. Hermenegildo
[pdf]
|
|
BYTECODE'07
|
An Efficient, Parametric Fixpoint Algorithm for Analysis of Java Bytecode
M. Méndez-Lojo, J. Navas, M. Hermenegildo
[pdf]
[BibTeX]
|
|
|
Invited talks and tutorials
|
|
|
|
HPCA'11
|
Parallelizing Irregular Applications through the Exploitation of Amorphous Data-Parallelism
K. Pingali, D. Nguyen, M. Méndez-Lojo
[webpage]
|
|
PLDI'10
|
Parallelizing Irregular Applications through the Exploitation of Amorphous Data-Parallelism
K. Pingali, M.Kulkarni, M. Méndez-Lojo, D. Nguyen
[webpage]
|
|
LFM'08
|
Safe Upper-bounds Inference of Energy Consumption for Java Bytecode Applications
J. Navas, M. Méndez-Lojo, M. Hermenegildo
The Sixth NASA Langley Formal Methods Workshop, May 2008.
|
|
SAC'06
|
Advanced Program Development Using Abstract Interpretation
M. Hermenegildo, E. Albert, P. Arenas, F. Bueno, M. Méndez-Lojo, G. Puebla
ACM SAC'06, ACM Symposium in Applied Computing (keynote address), Apr 2006
|
|
|
Technical Reports
|
|
Amorphous Data-parallelism in Irregular Algorithms
K. Pingali, M. Kulkarni, D. Nguyen, M. Burtscher, M. Méndez-Lojo, D. Prountzos, X.Sui, Z. Zhong
The University of Texas at Austin, TR-09-05, Feb 2009.
[pdf]
|
|
Technical reports prior to 2009 can be found here.
|
|
Doctoral Dissertation
|
|
A Generic Approach to Bytecode Analysis
M. Méndez-Lojo
University of New Mexico (UNM), Department of Computer Science, Aug 2008.
[pdf]
|
|
|
Ranked DBLP
|
| DBLP is a priceless source of information for Computer Science researchers. However, venues in the "Refine by Venue" box are not ranked, and sometimes we just do not know how good they are. My small contribution to solve this problem is a Greasemonkey user script that attaches to every conference its position in the Citeseer and CORE rankings. Install the script by clicking on it [requires Firefox and Greasemonkey]. |
|
|
