1. A Runtime Reconfigurable Design of Compute-in-Memory–Based Hardware Accelerator for Deep Learning Inference
- Author
-
Shimeng Yu, Shanshi Huang, Xiaochen Peng, Anni Lu, and Yandong Luo
- Subjects
010302 applied physics ,business.industry ,Computer science ,Computation ,Deep learning ,Inference ,02 engineering and technology ,01 natural sciences ,Computer Graphics and Computer-Aided Design ,Convolutional neural network ,020202 computer hardware & architecture ,Computer Science Applications ,Computer architecture ,Application-specific integrated circuit ,Memory wall ,0103 physical sciences ,0202 electrical engineering, electronic engineering, information engineering ,Hardware acceleration ,Artificial intelligence ,Electrical and Electronic Engineering ,business - Abstract
Compute-in-memory (CIM) is an attractive solution to address the “memory wall” challenges for the extensive computation in deep learning hardware accelerators. For custom ASIC design, a specific chip instance is restricted to a specific network during runtime. However, the development cycle of the hardware is normally far behind the emergence of new algorithms. Although some of the reported CIM-based architectures can adapt to different deep neural network (DNN) models, few details about the dataflow or control were disclosed to enable such an assumption. Instruction set architecture (ISA) could support high flexibility, but its complexity would be an obstacle to efficiency. In this article, a runtime reconfigurable design methodology of CIM-based accelerators is proposed to support a class of convolutional neural networks running on one prefabricated chip instance with ASIC-like efficiency. First, several design aspects are investigated: (1) the reconfigurable weight mapping method; (2) the input side of data transmission, mainly about the weight reloading; and (3) the output side of data processing, mainly about the reconfigurable accumulation. Then, a system-level performance benchmark is performed for the inference of different DNN models, such as VGG-8 on a CIFAR-10 dataset and AlexNet GoogLeNet, ResNet-18, and DenseNet-121 on an ImageNet dataset to measure the trade-offs between runtime reconfigurability, chip area, memory utilization, throughput, and energy efficiency.
- Published
- 2021