1. 首都师范大学 机构 信息工程学院,北京,102248
2. 电子系统可靠性技术北京市重点实验室 机构,北京,100048
3. 中国科学院 机构 计算技术研究所,北京,100080
4. 北京成像理论与技术高精尖创新中心,北京,100048
纸质出版:2019
移动端阅览
梁伟伟, 王晶, 王旭, 刘鑫培, 张伟功. 体系结构特征参数指导的近似计算技术[J]. 微电子学与计算机, 2019, 36(12): 72-77.
LIANG Wei-wei, WANG Jing, WANG Xu, et al. Architecture-parameters-guided approximate computing technique[J]. 2019, 36(12): 72-77.
能耗是制约微处理器发展的瓶颈之一
近似计算技术能够放松对精度的要求提高能效性.不同的应用程序对精度的需求各不相同
但现有技术选择近似程度时没有考虑程序需求
因而无法保证达到最优的能效性.针对上述问题本文提出了体系结构特征参数指导的近似计算技术
建立了体系结构参数同近似敏感度之间的联系
通过机器学习方法选择对近似影响较大的参数
利用分类决策树建立区分程序近似敏感度的分析模型
并基于分析结果指导近似程度的选择.实验结果显示
本文的方法能够为不同程序选择合适的近似程度
实现计算精度和能效性的折中优化.
This paper proposes an approximate computing technique for the guidance of architectural characteristic parameters. The technique analyzes the relationship between architectural parameters and approximate sensitive features. The parameters with great influence on approximate computing are selected by machine learning method.The classification and regression tree is used to analyze the approximate sensitivity of application and select the approximate degree according to the analysis results. Experimental results show that the method proposed by the paper can select the appropriate approximate degree for different application
and achieve the tradeoff optimization between computational accuracy and energy efficiency.
GRIGORIAN B, REINMAN G. Dynamically adaptive and reliable approximate computing using light-weight error analysis[C]// Adaptive Hardware and Systems. UK, Leicester: IEEE, 2014: 248-255.
CHIPPA V K, CHAKRADHAR S T, ROY K, et al. Analysis and characterization of inherent application resilience for approximate computing[C]// Proceedings of the 50th Annual Design Automation Conference. USA, Texas: ACM, 2013: 1-9.
HASHEMI S, ANTHONY N, TANN H, et al. Understanding the impact of precision quantization on the accuracy and energy of neural networks[C]//Proceedings of the Conference on Design, Automation % Test in Europe. Switzerland, Lausanne: EDAA, 2017: 1478-1483.
VENKATAGIRI R, MAHMOUD A, HARI S K S, et al. Approxilyzer: Towards a systematic framework for instruction-level approximate computing and its application to hardware resiliency[C]// The 49th Annual IEEE/ACM International Symposium on Microarchitecture. Taiwan, Taipei: IEEE, 2016: 1-14.
KAHNG A B, KANG S. Accuracy-configurable adder for approximate arithmetic designs[C]// Proceedings of the 49th Annual Design Automation Conference. USA, California: ACM, 2012: 820-825.
王珍珍, 杨志涛, 王晶, 等.基于BCH算法的高速缓存纠检错方案研究[J].微电子学与计算机, 2016, 33(11):124-128. DOI: 124
张迅珍, 梁青, 李涛.RV32I控制单元设计与实现[J].微电子学与计算机, 2018, 35(3):74-78. DOI: 74
ESMAEILZADEH H, SAMPSON A, CEZE L, et al. Neural acceleration for general-Purpose approximate programs[J]. IEEE Micro, 2013, 33(3):16-27. DOI: 16
0
浏览量
10
下载量
0
CSCD
关联资源
相关文章
相关作者
相关机构
京公网安备11010602201714号