arXiv subjects

Mei Wen

Publications and source records attributed to Mei Wen.

4 recordsLinked to original sources

arXiv2022

BP-Im2col: Implicit Im2col Supporting AI Backpropagation on Systolic Arrays

State-of-the-art systolic array-based accelerators adopt the traditional im2col algorithm to accelerate the inference of convolutional layers. However, traditional im2col cannot efficiently support AI backpropagation. Backpropagation in convolutional layers involves performing transposed convolution and dilated convolution, which usually introduces plenty of zero-spaces into the feature map or kernel. The zero-space data reorganization interfere with the continuity of training and incur additional and non-negligible overhead in terms of off- and on-chip storage, access and performance. Since countermeasures for backpropagation are rarely proposed, we propose BP-im2col, a novel im2col algorithm for AI backpropagation, and implement it in RTL on a TPU-like accelerator. Experiments on TPU-like accelerator indicate that BP-im2col reduces the backpropagation runtime by 34.9% on average, and reduces the bandwidth of off-chip memory and on-chip buffers by at least 22.7% and 70.6% respectively, over a baseline accelerator adopting the traditional im2col. It further reduces the additional storage overhead in the backpropagation process by at least 74.78%.

cs.AR↗

arXiv2019

Towards a Uniform Architecture for the Efficient Implementation of 2D and 3D Deconvolutional Neural Networks on FPGAs

Three-dimensional deconvolution is widely used in many computer vision applications. However, most previous works have only focused on accelerating 2D deconvolutional neural networks (DCNNs) on FPGAs, while the acceleration of 3D DCNNs has not been studied in depth as they have higher computational complexity and sparsity than 2D DCNNs. In this paper, we focus on the acceleration of both 2D and 3D DCNNs on FPGAs by proposing efficient schemes for mapping 2D and 3D DCNNs on a uniform architecture. By implementing our design on the Xilinx VC709 platform for four real-life 2D and 3D DCNNs, we can achieve up to 3.0 TOPS with high hardware efficiency. Comparisons with CPU and GPU solutions demonstrate that we can achieve an improvement of up to 63.3X in throughput relative to a CPU solution and an improvement of up to 8.3X in energy efficiency compared to a GPU solution.

cs.DC↗

arXiv2019

Metaflow: A DAG-Based Network Abstraction for Distributed Applications

In the past decade, increasingly network scheduling techniques have been proposed to boost the distributed application performance. Flow-level metrics, such as flow completion time (FCT), are based on the abstraction of flows yet they cannot capture the semantics of communication in a cluster application. Being aware of this problem, coflow is proposed as a new network abstraction. However, it is insufficient to reveal the dependencies between computation and communication. As a result, the real application performance can be hurt, especially in the absence of hard barriers. Based on the computation DAG of the application, we propose an expressive abstraction namely metaflow that resides in the middle of the two extreme points of flows and coflows. Evaluation results show that metaflow-based scheduling can outperform the coflow-based algorithm by 1.78x.

cs.NI↗

arXiv2018

Towards a Multi-array Architecture for Accelerating Large-scale Matrix Multiplication on FPGAs

Large-scale floating-point matrix multiplication is a fundamental kernel in many scientific and engineering applications. Most existing work only focus on accelerating matrix multiplication on FPGA by adopting a linear systolic array. This paper towards the extension of this architecture by proposing a scalable and highly configurable multi-array architecture. In addition, we propose a work-stealing scheme to ensure the equality in the workload partition among multiple linear arrays. Furthermore, an analytical model is developed to determine the optimal design parameters. Experiments on a real-life convolutional neural network (CNN) show that we can obtain the optimal extension of the linear array architecture.

cs.AR↗