[Colloq] =?Windows-1252?Q?Colloquium_Talk:_Friday, _October_18_ at _11_am, _366_WVH/_I-?= Ting Angelia Lee “Advances in Determinacy Race Detection for Task-Parallel Code”

[Khoury Academic Affairs]

Colloquium Talk: Friday, October 18 @ 11 am, 366 WVH/ I-Ting Angelia Lee “Advances in Determinacy Race Detection for Task-Parallel Code”

Speaker:  I-Ting Angelina Lee, Washington University in St. Louis

Title:  “Advances in Determinacy Race Detection for Task-Parallel Code”

Time: Friday, October 18, 11 AM

Place: 366 WVH

Abstract:
The widespread deployment of multicore platforms --- from personal computers to mobile devices to hardware for rent on the cloud --- has made it critical to develop simple approaches to programming them.  In this talk, I will discuss the progress that we made in addressing some of the challenges that arise in multicore programming.

I will focus on task parallelism, a programming model designed to simplify the job of writing parallel code that can utilize the multicore hardware efficiently.  With task parallelism, the programmer expresses the logical parallelism of the computation using high-level parallel control constructs, and lets the the underlying runtime system automates the necessary scheduling and synchronizations.

Even with task parallelism, writing correct and efficient parallel code can still be challenging.  One of the challenges is to deal with determinacy races, which occur when logically parallel parts of the computation access the same memory location and at least one of the accesses is a write.  Determinacy races are generally bugs in the program since they lead to non-deterministic program behavior --- different schedules of the program can lead to different results.  Moreover, they are difficult to detect and debug, since a race may manifest itself in one run but not in another.

In this talk, I will discuss our work on supporting efficient determinacy race detection for task-parallel code.  I will also briefly discuss how this work fits into my overall research agenda to simplify multicore programming.  I take the approach of tackling the problem from multiple perspectives: designing programming models, developing system support, and building an ecosystem of productivity tool supports around the model.  I will use our work on determinacy race detection as an example to illustrate that working from multiple perspectives can be synergistic and lead to results that are difficult to obtain otherwise.  If time permits, I will also briefly discuss some of my other ongoing work.

Bio:
I-Ting Angelina Lee is an assistant professor in the Computer Science and Engineering department in Washington University in St. Louis.  Her research agenda is to make multicore programming accessible for everyone, so that every programmer, particularly the non-experts, can rapidly develop high performance software that takes advantage of commodity multicore hardware.  To that end, she is interested in many aspects of multicore computing, including designing programming models and linguistic constructs to simplify multicore programming, developing runtime mechanisms and scheduling algorithms to enable parallel code to execute efficiently, and building productivity tools to aid debugging and performance engineering of parallel code.  She obtained her Ph.D. from Massachusetts Institute of Technology under the supervision of Professor Charles Leiserson.



Host: Rajmohan Rajaraman