First Friday Lunch with Julian Shun: High-Performance Frameworks for Static and Streaming Graph Processing

Speaker

Julian Shun
Assistant Professor, MIT EECS
There has been significant interest in high-performance graphprocessing due to their applications in many domains, including socialnetwork and Web analytics, machine learning, biology, and physicalsimulations. However, writing efficient parallel graph programs forprocessing the large-scale graphs available today can be verydifficult and time consuming, and therefore it is important to havetools that make the task easier. This talk will cover our recent workon high-level frameworks for parallel graph processing, for bothstatic graphs and streaming graphs. I will first present GraphIt, adomain-specific language that separates algorithm logic fromperformance optimizations to achieve high performance across differentstatic graph algorithms and datasets. I will then present Aspen, aframework for processing streaming graphs that introduces a newpurely-functional compressed tree data structure to enable graphqueries and updates to be performed concurrently with low latency. Oursolutions provide high-level programming interfaces that simplify thetask of writing parallel graph programs, while achieving highperformance at the same time.

About First Friday Lunches
First Friday Lunches are informal gatherings open to Alliances members and the CSAIL community, who would like to come by the lab for a discussion on a current project. Most lunches will feature a Faculty Researcher but a few will feature a PhD student, post-doc or Research Scientist. If you are not local but still want to participate, call in numbers as well as skype/webex connections will be made available. Please register for the lunch and indicate that you would like to participate remotely.

About Julian Shun
Julian Shun is the Douglas T. Ross Career Development AssistantProfessor of Software Technology in EECS and CSAIL at MIT. Prior tocoming to MIT, he was a Miller Research Fellow at UC Berkeley. Hereceived his Ph.D. from Carnegie Mellon University and his B.A. fromUC Berkeley. His research focuses on the theory and practice ofparallel algorithms and programming frameworks. He has received theNSF CAREER Award, DOE Early Career Award, ACM Doctoral DissertationAward, CMU School of Computer Science Doctoral Dissertation Award,Facebook Graduate Fellowship, best paper awards at PLDI and SPAA, andbest student paper award at DCC.