SiMul: An Algorithm-Driven Approximate Multiplier Design for Machine Learning
SiMul: An Algorithm-Driven Approximate Multiplier Design for Machine Learning
复制标题
DOI：
10.1109/mm.2018.043191125
复制
发表时间：
2018-07
期刊：
影响因子：
3.6
通讯作者：
Zhenhong Liu;A. Yazdanbakhsh;Taejoon Park;H. Esmaeilzadeh;N. Kim
中科院分区：
文献类型：
--
作者：
Zhenhong Liu;A. Yazdanbakhsh;Taejoon Park;H. Esmaeilzadeh;N. Kim
在能量受限的计算设备上支持各种机器学习（ML）算法的需求稳步增长。在本文中，我们提出了一个近似乘法器，这是各种ML加速器中的关键硬件组件。被称为SiMul，我们的近似乘法器具有用户控制的精度，利用ML算法的共同特点。SiMul支持在运行时的计算精度和能耗之间进行权衡，在满足所需推理精度要求的同时降低加速器的能耗。与精确乘法器相比，SiMul将乘法的能效提高了11.6倍至3.2倍，同时实现了81.7%至98.5%的乘法精度（三个不同应用程序的推理准确率分别为96.0%、97.8%和97.7%，而基线推理准确率为98.3%、99.0%，和97.7%使用精确的乘数）。使用我们的乘法器实现的神经加速器可以提供比使用精确乘法器实现的神经加速器高1.7倍（高达2.1倍）的能量效率，并且对各种应用的输出精度的影响可以忽略不计。
The need to support various machine learning (ML) algorithms on energy-constrained computing devices has steadily grown. In this article, we propose an approximate multiplier, which is a key hardware component in various ML accelerators. Dubbed SiMul, our approximate multiplier features user-controlled precision that exploits the common characteristics of ML algorithms. SiMul supports a tradeoff between compute precision and energy consumption at runtime, reducing the energy consumption of the accelerator while satisfying a desired inference accuracy requirement. Compared with a precise multiplier, SiMul improves the energy efficiency of multiplication by 11.6x to 3.2x while achieving 81.7-percent to 98.5-percent precision for individual multiplication operations (96.0-, 97.8-, and 97.7-percent inference accuracy for three distinct applications, respectively, compared to the baseline inference accuracy of 98.3, 99.0, and 97.7 percent using precise multipliers). A neural accelerator implemented with our multiplier can provide 1.7x (up to 2.1x) higher energy efficiency over one implemented with the precise multiplier with a negligible impact on the accuracy of the output for various applications.