PPoPP 2019
Sat 16 - Wed 20 February 2019 Washington, DC, United States
Tue 19 Feb 2019 14:25 - 14:50 at Salon 12/13 - Session 7: Scheduling Chair(s): Jidong Zhai

The use of futures provides a flexible way to express parallelism and can generate arbitrary dependences among parallel subcomputations. The additional flexibility that futures provide comes with a cost, however. When scheduled using classic work stealing, a program with futures, compared to a program that uses only fork-join parallelism, can incur a much higher number of ``deviations,'' a metric for evaluating the performance of parallel executions. All prior works assume a parsimonious work-stealing scheduler, however, where a worker thread (surrogate of a processor) steals work only when its local deque becomes empty.

In this work, we investigate an alternative scheduling approach, called ProWS, where the workers perform proactive work stealing when handling future operations. We show that ProWS, for programs that use futures, can provide provably efficient execution time and equal or better bounds on the number of deviations compared to classic parsimonious work stealing. Given a computation with $T_1$ work and $T_\infty$ span, ProWS executes the computation on $P$ processors in expected time $O(T_1 / P + T_\infty \lg P)$, with an additional $\lg P$ overhead on the span term compared to the parsimonious variant. For structured use of futures, where each future is single touch with no race on the future handle, the algorithm incurs $O(P T_\infty^2)$ deviations, matching that of the parsimonious variant. For general use of futures, the algorithm incurs $O(m_k T_\infty + P T_\infty \lg P)$ deviations, where $m_k$ is the maximum number of future touches that are logically parallel. Compared to the bound for the parsimonious variant, $O(k T_\infty + P T_\infty)$, with $k$ being the total number of touches in the entire computation, this bound is better assuming $m_k = \Omega(P \lg P)$ and is smaller than $k$, which holds true for all the benchmarks we examined.

Tue 19 Feb

Displayed time zone: Guadalajara, Mexico City, Monterrey change

14:00 - 15:15
Session 7: SchedulingMain Conference at Salon 12/13
Chair(s): Jidong Zhai Tsinghua University
14:00
25m
Talk
Semantics-Aware Scheduling Policies for Synchronization Determinism
Main Conference
Qi Zhao North Carolina State University, Zhengyi Qiu North Carolina State University, Guoliang Jin North Carolina State University
DOI
14:25
25m
Talk
Proactive Work Stealing for Futures
Main Conference
Kyle Singer Washington University in St. Louis, Yifan Xu Washington University in St. Louis, I-Ting Angelina Lee Washington University in St. Louis
DOI
14:50
25m
Talk
A Round-Efficient Distributed Betweenness Centrality Algorithm
Main Conference
Loc Hoang University of Texas at Austin, USA, Matteo Pontecorvi Nokia Bell Labs, Roshan Dathathri University of Texas at Austin, USA, Gurbinder Gill University of Texas at Austin, USA, Bozhi You Xi'an Jiaotong University, Keshav Pingali University of Texas at Austin, USA, Vijaya Ramachandran University of Texas at Austin
DOI