Energy-Aware GPU Programming at Source-Code Levels  

作　　者：Changyou Zhang Kun Huang Xiang Cui Yifeng Chen 

机构地区：[1]Key Laboratory of High Confidence Software Technologies(Peking University)School of Electronics Engineering and Computer Science,Peking University,Beijing 100871,China

出　　处：《Tsinghua Science and Technology》2012年第3期278-286,共9页清华大学学报（自然科学版（英文版）

基　　金：Supported by the National Natural Science Foundation of China (No. 61170053);the Natural Science Foundation of Beijing (No. 4112027);the China HGJ Significant Project (No. 2009ZX01036-001-002-4)

摘　　要：To enhance the energy efficiency and performance of algorithms with Graphics Processing Unit （GPU） accelerators in source-code development, we consider the power efficiency based on data transfer bandwidth and power consumption in key situations. First, a set of primitives is abstracted from program statements. Then, data transfer bandwidth and power consumption in different granularity sizes are consid- ered and mapped into proper primitives. With these mappings, a programmer can intuitively determine the power efficiency and performance in different running states of a thread. Finally, this intuition enables the programmer to tune the algorithm in order to achieve the best energy efficiency and performance. Using these power-aware principles, two Fast Fourier Transform （FFT） methods are compared. The mapping be- tween power consumption and primitives is helpful for algorithm tuning in source-code levels.To enhance the energy efficiency and performance of algorithms with Graphics Processing Unit （GPU） accelerators in source-code development, we consider the power efficiency based on data transfer bandwidth and power consumption in key situations. First, a set of primitives is abstracted from program statements. Then, data transfer bandwidth and power consumption in different granularity sizes are consid- ered and mapped into proper primitives. With these mappings, a programmer can intuitively determine the power efficiency and performance in different running states of a thread. Finally, this intuition enables the programmer to tune the algorithm in order to achieve the best energy efficiency and performance. Using these power-aware principles, two Fast Fourier Transform （FFT） methods are compared. The mapping be- tween power consumption and primitives is helpful for algorithm tuning in source-code levels.

关 键 词：GPU POWER-AWARE source-code PRIMITIVE 

分 类 号：TP316.81[自动化与计算机技术—计算机软件与理论] TP334.7[自动化与计算机技术—计算机科学与技术]