Semi-automatic restructuring of offloadable tasks for many-core accelerators

Author

Ravi, Nishkam ; Yi Yang ; Tao Bao ; Chakradhar, Srimat

Author_Institution

NEC Labs. America, Princeton, NJ, USA

fYear

2013

fDate

17-22 Nov. 2013

Firstpage

1

Lastpage

12

Abstract

Work division between the processor and accelerator is a common theme in modern heterogenous computing. Recent efforts (such as LEO and OpenAcc) provide directives that allow the developer to mark code regions in the original application from which offloadable tasks can be generated by the compiler. Auto-tuners and runtime schedulers work with the options (i.e., offloadable tasks) generated at compile time, which is limited by the directives specified by the developer. There is no provision for offload restructuring.

Keywords

multiprocessing systems; parallel architectures; parallel languages; processor scheduling; program compilers; CPU; EOC; Intel Xeon Phi architecture; LEO; MIC architecture; NAS parallel benchmarks; SpecOMP; accelerator memory; auto-tuners; code regions; directive-based languages; elastic offload compiler; heterogeneous computing; language extensions for offload; many-core accelerators; offloadable tasks; parallel loop; processor; relaxed semantics; runtime schedulers; semiautomatic restructuring; sub-offload; super-offload; Coprocessors; Data transfer; Indexes; Low earth orbit satellites; Microwave integrated circuits; Runtime; Semantics;

fLanguage

English

Publisher

ieee

Conference_Titel

High Performance Computing, Networking, Storage and Analysis (SC), 2013 International Conference for

Conference_Location

Denver, CO

Print_ISBN

978-1-4503-2378-9

Type

conf

DOI

10.1145/2503210.2503285

Filename

6877445