Hardware Acceleration of a Generalized Fast 2-D Convolution Method for Deep Neural Networks

Abstract

The hardware acceleration of Deep Neural Networks (DNN) is a highly effective and viable solution for running them on mobile devices. The power of DNNs is now available at the edge in a compact and power-efficient form factor with the aid of hardware acceleration. In this paper, we introduce an architecture that uses a generalized method called Single Partial Product 2-Dimensional Convolution (SPP2D Convolution) which calculates a 2-D convolution in a fast and expedient manner. We demonstrate that the SPP2D architecture prevents the re-fetching of input weights for the calculation of partial products , and it can calculate the output of any input size and kernel with low latency and high throughput compared to other popular techniques. SPP2D based architecture can reduce the memory access and execution time related to input reuse by at least three times in comparison with the work done in [1] and approximately nine times that of the standard sliding window approach. We have implemented the generalized SPP2D architecture on the Xilinx KC705 Kintex-7 evaluation board to illustrate that the new SPP2D algorithm is well-suited for the hardware acceleration of DNNs. We implemented LeNet-5 [2] and VGGNet-16 [3] using the SPP2D architecture. We demonstrate that the SPP2D based LeNet-5 has a high throughput of 5 GOP/s and 14.8 GOP/s/W and 42 GOP/s/W for the convolution operation using the SPP2D IP. Our LeNet-5 design achieves a similar throughput to [4] however using 3.3× fewer DSPs and an even smaller memory and lookup table (LUT) footprint. The SPP2D based VGGNet-16 network has a latency of 91.3 ms which is 79%, 97%, 17% and 95% less than contemporary designs [1], [5]-[7] respectively, while running at a low power of 298 mW which is similar to the power level of these designs. The total processing time of our design with a parallelism factor of nine is 3.93 secs and it is 70% less than that in [1] and 24% less than that in [7]. The SPP2D based LeNet-5 and VGGNet-16 accelerators provide a low-latency design with reduced memory access thus leading to a low-power design.As a result, SPP2D convolution is very well suited for hardware acceleration of DNNs.

High-Speed CNN Accelerator SoC Design Based on a Flexible Diagonal Cyclic Array

“…The acceleration of CNN model inference for object detection is discussed in more detail, with a focus on FPGAbased implementations. Thus, various hardware architecture approaches and optimization methods are explored to examine their impact on throughput and accuracy [17][18][19][20][21][22][23][24][25][26].…”

Optimization of Control Method of Construction Machinery Manipulator Based on Neural Network

“…At the same time, considering that neural network structure, system performance and other factors have a large impact on data processing speed and accuracy, analog methods can be used for control. The idea of the neural network algorithm is to process the information in the complex system, transform it into a simple function, and then predict and classify it [16][17].…”

Relationship between Marine Resources growth and Marine Economy Based on Neural Network

“…Moreover, with the growth of population and the improvement of productivity, people pay more and more attention to a series of influencing factors such as environmental problems brought by the process of economic socialization. Therefore, the growth and utilization of marine organisms can not only improve the ecological environment, but also promote sustainable economic growth [13][14].…”