摘要: In response to the current requirements of energy efficiency and high performance in computing systems, architects have turned towards customization. General purpose however remains a challenge as processors must adhere variety applications, on-chip resources, increased without solely relying on transistor scaling additional cache levels. For this reason, concept Reconfigurable Computing Unit (RCU) been proposed which redesign conventional processor microarchitectural architectural level. RCUs are extended work support multi-task workload using OmpSs, where task instruction placement algorithms thoroughly assessed for effects efficiency. Experimental results demonstrate that single RCU with double engine configuration is able exceed single-core average by 1.48x achieve/exceed dual-core performance. The various inter-and intra-task tested also display up 16.7% 23% fluctuation efficiency, respectively, depending method combination employed.
sci-hub.se 参考文章(5)
Anita Tino, Kaamran Raahemifar, Addressing processor back-end issues with RCUs adaptive hardware and systems. pp. 1- 8 ,(2015) , 10.1109/AHS.2015.7231173
Michael J. Flynn, Oskar Mencer, Veljko Milutinovic, Goran Rakocevic, Per Stenstrom, Roman Trobec, Mateo Valero, Moving from petaflops to petadata Communications of the ACM. ,vol. 56, pp. 39- 42 ,(2013) , 10.1145/2447976.2447989
Shekhar Borkar, Andrew A. Chien, The future of microprocessors Communications of The ACM. ,vol. 54, pp. 67- 77 ,(2011) , 10.1145/1941487.1941507
Tamer Dallou, Ben Juurlink, Hardware-Based Task Dependency Resolution for the StarSs Programming Model 2012 41st International Conference on Parallel Processing Workshops. pp. 367- 374 ,(2012) , 10.1109/ICPPW.2012.53
Hadi Esmaeilzadeh, Adrian Sampson, Luis Ceze, Doug Burger, Neural Acceleration for General-Purpose Approximate Programs international symposium on microarchitecture. ,vol. 33, pp. 449- 460 ,(2012) , 10.1109/MICRO.2012.48