BBS: Bi-directional Bit-level Sparsity for Deep Learning Acceleration

TL;DR

This paper introduces Bidirectional Bit sparsity (BBS), a novel method that leverages symmetrical bit pruning and a hardware accelerator to significantly improve deep learning efficiency, reducing model size, increasing speed, and saving energy.

Contribution

The paper proposes BBS, a new symmetric bit pruning technique combined with a hardware accelerator, enhancing bit-level sparsity and efficiency in deep neural network acceleration.

Findings

Achieves 1.66× reduction in model size with minimal accuracy loss

Provides up to 3.03× speedup over prior accelerators

Saves 2.44× energy compared to existing DNN hardware

Abstract

Bit-level sparsity methods skip ineffectual zero-bit operations and are typically applicable within bit-serial deep learning accelerators. This type of sparsity at the bit-level is especially interesting because it is both orthogonal and compatible with other deep neural network (DNN) efficiency methods such as quantization and pruning. In this work, we improve the practicality and efficiency of bitlevel sparsity through a novel algorithmic bit-pruning, averaging, and compression method, and a co-designed efficient bit-serial hardware accelerator. On the algorithmic side, we introduce bidirectional bit sparsity (BBS). The key insight of BBS is that we can leverage bit sparsity in a symmetrical way to prune either zero-bits or one-bits. This significantly improves the load balance of bit-serial computing and guarantees the level of sparsity to be more than 50%. On top of BBS, we further propose two bit-level binary pruning methods that require no retraining, and can be seamlessly applied to quantized DNNs. Combining binary pruning with a new tensor encoding scheme, BBS can both skip computation and reduce the memory footprint associated with bi-directional sparse bit columns. On the hardware side, we demonstrate the potential of BBS through BitVert, a bitserial architecture with an efficient PE design to accelerate DNNs with low overhead, exploiting our proposed binary pruning. Evaluation on seven representative DNN models shows that our approach achieves: (1) on average 1.66$\times$ reduction in model sizewith negligible accuracy loss of < 0.5%; (2) up to 3.03$\times$ speedupand 2.44$\times$ energy saving compared to prior DNN accelerators.