Flex-SFU：通过非均匀分段逼近加速DNN激活函数
Flex-SFU: Accelerating DNN Activation Functions by Non-Uniform Piecewise Approximation
AI总结：
提出Flex-SFU轻量级硬件加速器，采用非均匀分段插值支持浮点，平均改善均方误差22.3倍，提升AI加速器端到端性能35.7%，面积功耗开销极小。
摘要：
现代DNN工作负载日益依赖由计算复杂操作组成的激活函数。这对当前针对卷积和矩阵-矩阵乘法优化的加速器构成了挑战。本工作提出了Flex-SFU，一种用于激活函数的轻量级硬件加速器，实现支持多种数据格式的非均匀分段插值。通过在地址解码单元中实现二叉树比较，支持非均匀分段和浮点数。提出了一种基于SGD的启发式优化算法，以找到降低均方误差的插值函数。得益于非均匀插值和浮点支持，与先前的分段线性插值方法相比，Flex-SFU平均实现了22.3倍的均方误差改善。对超过700个计算机视觉和自然语言处理模型的评估表明，Flex-SFU平均可将最先进的AI硬件加速器的端到端性能提升35.7%，在使用32个分段时实现高达3.3倍的加速，且对模型准确率的影响可忽略不计，相对于基线向量处理单元仅引入5.9%的面积开销和0.8%的功耗开销。
英文摘要：
Modern DNN workloads increasingly rely on activation functions consisting of computationally complex operations. This poses a challenge to current accelerators optimized for convolutions and matrix-matrix multiplications. This work presents Flex-SFU, a lightweight hardware accelerator for activation functions implementing non-uniform piecewise interpolation supporting multiple data formats. Non-Uniform segments and floating-point numbers are enabled by implementing a binary-tree comparison within the address decoding unit. An SGD-based optimization algorithm with heuristics is proposed to find the interpolation function reducing the mean squared error. Thanks to non-uniform interpolation and floating-point support, Flex-SFU achieves on average 22.3x better mean squared error compared to previous piecewise linear interpolation approaches. The evaluation with more than 700 computer vision and natural language processing models shows that Flex-SFU can, on average, improve the end-to-end performance of state-of-the-art AI hardware accelerators by 35.7%, achieving up to 3.3x speedup with negligible impact in the models' accuracy when using 32 segments, and only introducing an area and power overhead of 5.9% and 0.8% relative to the baseline vector processing unit.
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。
