Title :
Limits of Parallelism and Boosting in Dim Silicon
Author :
Pinckney, N. ; Dreslinski, Ronald G. ; Sewell, K. ; Fick, David ; Mudge, Trevor ; Sylvester, Dennis ; Blaauw, D.
Author_Institution :
Univ. of Michigan, Ann Arbor, MI, USA
Abstract :
Supply-voltage scaling has stagnated in recent technology nodes, leading to so-called dark silicon. To increase overall chip multiprocessor (CMP) performance, it is necessary to improve the energy efficiency of individual tasks so that more tasks can be executed simultaneously within thermal limits. In this article, the authors investigate the limit of voltage scaling together with task parallelization to maintain task completion latency while reducing energy consumption. Additionally, they examine improvements in energy efficiency and parallelism when serial portions of code can be overcome through quickly boosting a core´s operating voltage. When accounting for parallelization overheads, minimum task energy is obtained at near-threshold supply voltages across six commercial technology nodes and provides 4× improvement in overall CMP performance. Boosting is most effective when the task is modestly parallelizable but not highly parallelizable.
Keywords :
energy conservation; energy consumption; microprocessor chips; multiprocessing systems; performance evaluation; power aware computing; CMP performance improvement; chip multiprocessor performance improvement; commercial technology nodes; dark silicon; dim silicon; energy consumption reduction; energy efficiency improvement; near-threshold supply voltages; operating voltage; parallelization overheads; supply-voltage scaling; task completion latency; task parallelization; thermal limits; Boosting; Energy efficiency; Logic gates; Parallel processing; Semiconductor device manufacture; Silicon; Transistors; Voltage control; energy-aware systems; low-power design;
Journal_Title :
Micro, IEEE