DocumentCode :
2313098
Title :
Parallelism through dynamic instrumentation at runtime
Author :
Yadav, Raj ; Singh, Moirangthem Dennis ; Mahajan, Neha
Author_Institution :
Digital Media Center, Samsung India Electron. Pvt. Ltd., Noida, India
fYear :
2010
fDate :
9-11 Feb. 2010
Firstpage :
321
Lastpage :
325
Abstract :
This paper presents a novel approach to achieve parallelism on multi-core systems out of the legacy software without recompilation. A profiler tool can be enhanced, from identifying the bottleneck areas, to analyzing the instruction set in bottleneck areas. As the instructions along with all data dependencies are available in the running program, heuristics can be applied to detect the candidates for instruction level parallelism. The serial regions can be regenerated into parallel regions for multiple cores using predefined OpenMP calls and instrument dynamically at runtime. We discuss the problems for parallelism 1) Identifying the parallel regions for parallelism from serial code 2) Detailed approach for generating code generation at runtime.
Keywords :
instruction sets; message passing; parallel programming; program compilers; software maintenance; system monitoring; OpenMP calls; code generation; data dependency; instruction level parallelism; instruction set analysis; legacy software; multicore system; profiler tool; runtime dynamic instrumentation; serial code; Concurrent computing; Data analysis; Hardware; Instruments; Machine learning; Memory management; Multicore processing; Parallel processing; Runtime; Auto Parallellism; Dynamic Instrumentation; Multi-core architecture; OpenMp; Runtime code generation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Machine Learning and Computing (ICMLC), 2010 Second International Conference on
Conference_Location :
Bangalore
Print_ISBN :
978-1-4244-6006-9
Electronic_ISBN :
978-1-4244-6007-6
Type :
conf
DOI :
10.1109/ICMLC.2010.58
Filename :
5460717
Link To Document :
بازگشت