Design and Performance Characterization of RADICAL-Pilot on Leadership-class Platforms
Authors:
Preprint
Abstract
Many extreme scale scientific applications have workloads comprised of a large number of individual high-performance tasks. The Pilot abstraction decouples workload specification, resource management, and task execution via job placeholders and late-binding. As such, suitable implementations of the Pilot abstraction can support the collective execution of large number of tasks on supercomputers. We introduce RADICAL-Pilot (RP) as a portable, modular and extensible pilot-enabled runtime system. We describe RP's design, architecture and implementation. We characterize its performance and show its ability to scalably execute workloads comprised of tens of thousands heterogeneous tasks on DOE and NSF leadership-class HPC platforms. Specifically, we investigate RP's weak/strong scaling with CPU/GPU, single/multi core, (non)MPI tasks and Python functions when using most of ORNL Summit and TACC Frontera. RADICAL-Pilot can be used stand-alone, as well as the runtime for third-party workflow systems.Keywords
All Related Versions
This publication has 16 references indexed in Scilit:
- Synapse: Synthetic application profiler and emulatorJournal of Computational Science, 2018
- SAGA: A standardized access layer to heterogeneous Distributed Computing InfrastructureSoftwareX, 2015
- FireWorks: a dynamic workflow system designed for high‐throughput applicationsConcurrency and Computation: Practice and Experience, 2015
- Pegasus, a workflow management system for science automationFuture Generation Computer Systems, 2015
- GWpilot: Enabling multi-level scheduling in distributed infrastructures with GridWay and pilot jobsFuture Generation Computer Systems, 2015
- Fast recovery of free energy landscapes via diffusion-map-directed molecular dynamicsPhysical Chemistry Chemical Physics, 2014
- JETS: Language and System Support for Many-Parallel-Task WorkflowsJournal of Grid Computing, 2013
- CernVM Co-Pilot: an Extensible Framework for Building Scalable Computing Infrastructures on the CloudJournal of Physics: Conference Series, 2012
- Pilot factory – a Condor-based system for scalable Pilot Job generation in the Panda WMS frameworkJournal of Physics: Conference Series, 2010
- Cilk: An Efficient Multithreaded Runtime SystemJournal of Parallel and Distributed Computing, 1996