arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2201.08022cs.ARcs.AI
收藏

HEAM:面向深度神经网络的高效近似乘法器优化

HEAM: High-Efficiency Approximate Multiplier Optimization for Deep Neural Networks

State Key Laboratory of ASIC & System, Fudan University(复旦大学专用集成电路与系统国家重点实验室)

Su Zheng, Zhen Li, Yao Lu, Jingbo Gao, Jide Zhang, Lingli Wang

AI总结:

提出一种根据操作数分布最小化平均误差的近似乘法器自动设计优化方法,在几乎不损失准确率的前提下显著降低DNN的面积、功耗和延迟。

摘要:

我们提出了一种近似乘法器自动设计的优化方法,该方法根据操作数分布最小化平均误差。在DNN中,我们的乘法器比最佳复现近似乘法器准确率高出50.24%,面积减小15.76%,功耗降低25.05%,延迟缩短3.50%。与精确乘法器相比,面积、功耗和延迟分别降低44.94%、47.63%和16.78%,准确率损失可忽略不计。采用该乘法器的DNN加速器模块面积减小18.70%,功耗降低9.99%。

英文摘要:

We propose an optimization method for the automatic design of approximate multipliers, which minimizes the average error according to the operand distributions. Our multiplier achieves up to 50.24% higher accuracy than the best reproduced approximate multiplier in DNNs, with 15.76% smaller area, 25.05% less power consumption, and 3.50% shorter delay. Compared with an exact multiplier, our multiplier reduces the area, power consumption, and delay by 44.94%, 47.63%, and 16.78%, respectively, with negligible accuracy losses. The tested DNN accelerator modules with our multiplier obtain up to 18.70% smaller area and 9.99% less power consumption than the original modules.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑