Scalable Work Stealing
Irregular and dynamic parallel applications pose significant challenges to achieving scalable performance on large-scale multicore clusters. These applications often require ongoing, dynamic load balancing in order to maintain efficiency. While effective at small scale, centralized load balancing schemes quickly become a bottleneck on large-scale clusters. Work stealing is a popular approach to distributed dynamic load balancing; however its performance on large-scale clusters is not well understood. Prior work on work stealing has largely focused on shared memory machines. In this work we investigate the design and scalability of work stealing on modern distributed memory systems. We demonstrate high efficiency and low overhead when scaling to 8,192 processors for three benchmark codes: a producer-consumer benchmark, the unbalanced tree search benchmark, and a multiresolution analysis kernel.
- Research Organization:
- Pacific Northwest National Laboratory (PNNL), Richland, WA (US)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC05-76RL01830
- OSTI ID:
- 986715
- Report Number(s):
- PNNL-SA-67261; KJ0402000
- Country of Publication:
- United States
- Language:
- English
Similar Records
Work stealing for GPU-accelerated parallel programs in a global address space framework: WORK STEALING ON GPU-ACCELERATED SYSTEMS
Steal Tree: Low-Overhead Tracing of Work Stealing Schedulers
Work stealing for GPU-accelerated parallel programs in a global address space framework
Journal Article
·
Tue Jan 05 23:00:00 EST 2016
· Concurrency and Computation. Practice and Experience
·
OSTI ID:1333989
Steal Tree: Low-Overhead Tracing of Work Stealing Schedulers
Conference
·
Sun Jun 16 00:00:00 EDT 2013
·
OSTI ID:1237828
Work stealing for GPU-accelerated parallel programs in a global address space framework
Journal Article
·
Tue Jan 05 23:00:00 EST 2016
· Concurrency and Computation. Practice and Experience
·
OSTI ID:1393474