Minimizing Power for Neural Network Training with Logarithm-Approximate Floating-Point Multiplier
Minimizing Power for Neural Network Training with Logarithm-Approximate Floating-Point Multiplier
复制标题
DOI：
10.1109/patmos.2019.8862162
复制
发表时间：
2019-07
期刊：
影响因子：
--
通讯作者：
TaiYu Cheng;Jaehoon Yu;M. Hashimoto
中科院分区：
文献类型：
--
作者：
TaiYu Cheng;Jaehoon Yu;M. Hashimoto
本文提出了一种神经网络（NN）训练引擎中的乘-累加（MAC）计算方法--我们实现的NN训练引擎的2-D分类数据集实现了10%的速度和2.5倍和2.3倍的效率提高功率和面积，分别。LAM还与传统的位宽缩放（BWS）高度兼容。当在四个测试数据集中应用BWS与LAM时，可以实现超过5.2倍的功率效率改进，而精度仅下降1%，其中2.3倍的改进来自LAM。
This paper proposes to adopt logarithm-approximate multiplier (LAM) for multiply-accumulate (MAC) computation in neural network (NN) training engine, where LAM approximates a floating-point multiplication as an addition resulting in smaller delay, fewer gates, and lower power consumption. Our implementation of NN training engine for a 2-D classification dataset achieves 10% speed-up and 2.5X and 2.3X efficiency improvement in power and area, respectively. LAM is also highly compatible with conventional bit-width scaling (BWS). When BWS is applied with LAM in four test datasets, more than 5.2X power efficiency improvement is achievable with only 1% accuracy degradation, where 2.3X improvement originates from LAM.