Title :
Interconnect compression and its benefits for multi-core systems
Author :
Liu, Jiangjiang ; Zhang, Jianyong
Author_Institution :
Deptartment of Comput. Sci., Lamar Univ., Beaumont, TX, USA
Abstract :
One of the major problems associated with multi-core systems is the performance and energy demand it places on the interconnect system. We presents an overall analysis of the compression possibility in the instructions and data communicated on interconnects for improving performance and reducing energy in multi-core systems. Our analysis shows great potential of interconnect compression. In our study, lower compression ratio means better compression. We find that up to 0.71 compression ratios can be obtained for instruction and data transferred on interconnect components and for integer and floating-point applications, up to 0.74 and 0.73 compression ratios can be achieved, respectively.
Keywords :
data communication; floating point arithmetic; integrated circuit interconnections; multiprocessing systems; compression ratio; data communication; data transfer; energy demand; floating-point applications; integer applications; interconnect components; interconnect compression; multicore systems; Benchmark testing; Entropy; Integrated circuit interconnections; Markov processes; Multicore processing; Power demand; Transistors;
Conference_Titel :
SOC Conference (SOCC), 2012 IEEE International
Conference_Location :
Niagara Falls, NY
Print_ISBN :
978-1-4673-1294-3
DOI :
10.1109/SOCC.2012.6398389
