Networked embedded systems typically leverage a collection of low-powerembedded systems (nodes) to collaboratively execute applications spanningdiverse application domains (e.g., video, image processing, communication,etc.) with diverse application requirements. The individual networked nodesmust operate under stringent constraints (e.g., energy, memory, etc.) andshould be specialized to meet varying application requirements in order toadhere to these constraints. Phase-based tuning specializes system tunableparameters to the varying runtime requirements of different execution phases tomeet optimization goals. Since the design space for tunable systems can be verylarge, one of the major challenges in phase-based tuning is determining thebest configuration for each phase without incurring significant tuning overhead(e.g., energy and/or performance) during design space exploration. In thispaper, we propose phase distance mapping, which directly determines the bestconfiguration for a phase, thereby eliminating design space exploration. Phasedistance mapping applies the correlation between the characteristics and bestconfiguration of a known phase to determine the best configuration of a newphase. Experimental results verify that our phase distance mapping approach,when applied to cache tuning, determines cache configurations within 1 % of theoptimal configurations on average and yields an energy delay product savings of27 % on average.
展开▼
