Write a Blog >>
Wed 21 Jun 2017 16:45 - 17:10 at Auditorium, Vertex Building - Analysis and Synthesis Chair(s): Anders Møller

In a precise data race detector, a race is detected only if the execution exhibits a real race. In such tools, every memory access from each thread is typically checked by a happens-before algorithm. What’s the optimal runtime performance of such tools? In this paper, we identify that a significant percentage of memory access checks in real-world program executions are often redundant: removing these checks neither affect the precision nor the capability of race detection. We show that if all such redundant checks were eliminated the performance of a state-of-the-art dynamic race detector, FastTrack, could be improved by 95%, reducing its runtime overhead from 77X to 3.4X on a collection of CPU intensive benchmarks running a large number of threads. We further develop a purely dynamic technique, ReX, that efficiently filters out redundant checks and apply it FastTrack. With ReX, the runtime performance of FastTrack is improved by 25% on average.

Wed 21 Jun

ecoop-2017-papers
15:30 - 17:10: ECOOP Research Papers - Analysis and Synthesis at Auditorium, Vertex Building
Chair(s): Anders MøllerAarhus University
ecoop-2017-papers15:30 - 15:55
Talk
Mikaël MayerEPFL, Switzerland, Jad HamzaLIAFA, Université Paris Diderot, Viktor KuncakEPFL, Switzerland
Link to publication Media Attached
ecoop-2017-papers15:55 - 16:20
Talk
Shiyou HuangTexas A&M University, Jeff HuangTexas A&M University
Link to publication Media Attached
ecoop-2017-papers16:20 - 16:45
Talk
Quentin StiévenartVrije Universiteit Brussel, Belgium, Jens NicolayVrije Universiteit Brussel, Belgium, Wolfgang De MeuterVrije Universiteit Brussel, Coen De RooverVrije Universiteit Brussel
Link to publication Pre-print Media Attached
ecoop-2017-papers16:45 - 17:10
Talk
Jeff HuangTexas A&M University, Arun Krishnakumar RajagopalanTexas A&M University
Link to publication Media Attached