DocumentCode :
1997055
Title :
Composing Multiple StarPU Applications over Heterogeneous Machines: A Supervised Approach
Author :
Hugo, Andra-Ecaterina ; Guermouche, Abdou ; Wacrenier, Pierre-Andre ; Namyst, Raymond
Author_Institution :
LaBRI, Univ. of Bordeaux, Talence, France
fYear :
2013
fDate :
20-24 May 2013
Firstpage :
1050
Lastpage :
1059
Abstract :
Enabling HPC applications to perform efficiently when invoking multiple parallel libraries simultaneously is a great challenge. Even if a single runtime system is used underneath, scheduling tasks or threads coming from different libraries over the same set of hardware resources introduces many issues, such as resource oversubscription, undesirable cache flushes or memory bus contention. This paper presents an extension of StarPU, a runtime system specifically designed for heterogeneous architectures, that allows multiple parallel codes to run concurrently with minimal interference. Such parallel codes run within scheduling contexts that provide confined execution environments which can be used to partition computing resources. Scheduling contexts can be dynamically resized to optimize the allocation of computing resources among concurrently running libraries. We introduce a hypervisor that automatically expands or shrinks contexts using feedback from the runtime system (e.g. resource utilization). We demonstrate the relevance of our approach using benchmarks invoking multiple high performance linear algebra kernels simultaneously on top of heterogeneous multicore machines. We show that our mechanism can dramatically improve the overall application run time (-34%), most notably by reducing the average cache miss ratio (-50%).
Keywords :
linear algebra; multiprocessing systems; parallel processing; resource allocation; scheduling; HPC applications; StarPU runtime system; application run time; average cache miss ratio; computing resource allocation; concurrently running libraries; hardware resources; heterogeneous architectures; heterogeneous multicore machines; high performance computing; high performance linear algebra kernels; hypervisor; multiple parallel codes; parallel libraries; resource utilization; scheduling contexts; single runtime system; task scheduling; thread scheduling; Context; Dynamic scheduling; Kernel; Libraries; Runtime; Virtual machine monitors; Composability; Runtime; StarPU;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Parallel and Distributed Processing Symposium Workshops & PhD Forum (IPDPSW), 2013 IEEE 27th International
Conference_Location :
Cambridge, MA
Print_ISBN :
978-0-7695-4979-8
Type :
conf
DOI :
10.1109/IPDPSW.2013.217
Filename :
6650990
Link To Document :
بازگشت