Title :
Underclocked Software Prefetching: More Cores, Less Energy
Author :
Kamruzzaman, Md ; Swanson, Steven ; Tullsen, Dean M.
Author_Institution :
Dept. of Comput. Sci. & Eng., Univ. of California, San Diego, La Jolla, CA, USA
Abstract :
Power consumption is a concern for helper-thread prefetching that uses extra cores to speed up the single-thread execution, because power consumption increases with each additional core. This article analyzes the impact of using power-saving techniques in the context of intercore prefetching (ICP), and shows that dynamic frequency scaling coupled with ICP is a more energy-efficient way to improve the single-thread execution.
Keywords :
cache storage; multi-threading; multiprocessing systems; ICP; dynamic frequency scaling; helper-thread prefetching; intercore prefetching; power consumption; power saving techniques; single-thread execution improvement; underclocked software prefetching; Energy efficiency; Iterative closest point algorithm; Multicore processing; Power demand; Power measurement; Prefetching; Voltage control; Energy efficiency; Iterative closest point algorithm; Multicore processing; Power demand; Power measurement; Prefetching; Voltage control; energy-efficient computing; frequency/voltage scaling; helper-thread prefetching; intercore prefetching; single-thread performance;
Journal_Title :
Micro, IEEE
