DAGuE: A Generic Distributed DAG Engine for High Performance Computing
Authors:
- George BosilcaGBGeorge Bosilca
- Aurelien BouteillerABAurelien Bouteiller
- Anthony DanalisADAnthony Danalis
- Thomas HeraultTHThomas Herault
- Pierre LemarinierPLPierre Lemarinier
- Jack DongarraJDJack Dongarra
- 1 May 2011
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 38 (15302075) , 1151-1158
Abstract
The frenetic development of the current architectures places a strain on the current state-of-the-art programming environments. Harnessing the full potential of such architectures has been a tremendous task for the whole scientific computing community. We present DAGuE a generic framework for architecture aware scheduling and management of micro-tasks on distributed many-core heterogeneous architectures. Applications we consider can be represented as a Direct Acyclic Graph of tasks with labeled edges designating data dependencies. DAGs are represented in a compact, problem-size independent format that can be queried on-demand to discover data dependencies, in a totally distributed fashion. DAGuE assigns computation threads to the cores, overlaps communications and computations and uses a dynamic, fully-distributed scheduler based on cache awareness, data-locality and task priority. We demonstrate the efficiency of our approach, using several micro-benchmarks to analyze the performance of different components of the framework, and a Linear Algebra factorization as a use case.Keywords
This publication has 17 references indexed in Scilit:
- hwloc: A Generic Framework for Managing Hardware Affinities in HPC Applications2010
- Dynamic task scheduling for linear algebra algorithms on distributed-memory multicore systems2009
- Distributed SBP Cholesky factorization algorithms with near-optimal schedulingACM Transactions on Mathematical Software, 2009
- A class of parallel tiled linear algebra algorithms for multicore architecturesParallel Computing, 2009
- Multi-threading and one-sided communication in parallel LU factorization2007
- The Impact of Multicore on Math Software2007
- Grid'5000: A Large Scale And Highly Reconfigurable Experimental Grid TestbedThe International Journal of High Performance Computing Applications, 2006
- Workflow Global Computing with YML2006
- Compact DAG representation and its symbolic schedulingJournal of Parallel and Distributed Computing, 2004
- AUTOMATIC PARALLELIZATION TECHNIQUES BASED ON COMPACT DAG EXTRACTION AND SYMBOLIC SCHEDULINGParallel Processing Letters, 2001