SPOTS: An Accelerator for Sparse Convolutional Networks Leveraging Systolic General Matrix-Matrix Multiplication

by   Mohammadreza Soltaniyeh, et al.
Rutgers University

This paper proposes a new hardware accelerator for sparse convolutional neural networks (CNNs) by building a hardware unit to perform the Image to Column (IM2COL) transformation of the input feature map coupled with a systolic array-based general matrix-matrix multiplication (GEMM) unit. Our design carefully overlaps the IM2COL transformation with the GEMM computation to maximize parallelism. We propose a novel design for the IM2COL unit that uses a set of distributed local memories connected by a ring network, which improves energy efficiency and latency by streaming the input feature map only once. We propose a tall systolic array for the GEMM unit while also providing the ability to organize it as multiple small GEMM units, which enables our design to handle a wide range of CNNs and their parameters. Further, our design improves performance by effectively mapping the sparse data to the hardware units by utilizing sparsity in both input feature maps and weights. Our prototype, SPOTS, is on average 1.74X faster than Eyeriss. It is also 78X, and 12X more energy-efficient when compared to CPU and GPU implementations, respectively.


page 1

page 8

page 9

page 10


SCNN: An Accelerator for Compressed-sparse Convolutional Neural Networks

Convolutional Neural Networks (CNNs) have emerged as a fundamental techn...

ArrayFlex: A Systolic Array Architecture with Configurable Transparent Pipelining

Convolutional Neural Networks (CNNs) are the state-of-the-art solution f...

Efficient Implementation of Multi-Channel Convolution in Monolithic 3D ReRAM Crossbar

Convolutional neural networks (CNNs) demonstrate promising accuracy in a...

Hardware-Efficient Template-Based Deep CNNs Accelerator Design

Acceleration of Convolutional Neural Network (CNN) on edge devices has r...

Systolic Tensor Array: An Efficient Structured-Sparse GEMM Accelerator for Mobile CNN Inference

Convolutional neural network (CNN) inference on mobile devices demands e...

ADC/DAC-Free Analog Acceleration of Deep Neural Networks with Frequency Transformation

The edge processing of deep neural networks (DNNs) is becoming increasin...

Please sign up or login with your details

Forgot password? Click here to reset