SCNN: An Accelerator for Compressed-sparse Convolutional Neural Networks
This work addresses the need for efficient CNN deployment in mobile devices like autonomous vehicles and cameras, representing an incremental advance in accelerator design.
The paper tackles the problem of improving performance and energy efficiency for convolutional neural networks (CNNs) on mobile platforms by introducing the SCNN accelerator architecture, which exploits sparsity from pruned weights and ReLU activations, achieving 2.7x performance and 2.3x energy improvements over dense accelerators.
Convolutional Neural Networks (CNNs) have emerged as a fundamental technology for machine learning. High performance and extreme energy efficiency are critical for deployments of CNNs in a wide range of situations, especially mobile platforms such as autonomous vehicles, cameras, and electronic personal assistants. This paper introduces the Sparse CNN (SCNN) accelerator architecture, which improves performance and energy efficiency by exploiting the zero-valued weights that stem from network pruning during training and zero-valued activations that arise from the common ReLU operator applied during inference. Specifically, SCNN employs a novel dataflow that enables maintaining the sparse weights and activations in a compressed encoding, which eliminates unnecessary data transfers and reduces storage requirements. Furthermore, the SCNN dataflow facilitates efficient delivery of those weights and activations to the multiplier array, where they are extensively reused. In addition, the accumulation of multiplication products are performed in a novel accumulator array. Our results show that on contemporary neural networks, SCNN can improve both performance and energy by a factor of 2.7x and 2.3x, respectively, over a comparably provisioned dense CNN accelerator.