ESACT: An End-to-End Sparse Accelerator for Compute-Intensive Transformers via Local Similarity

TL;DR

This paper introduces ESACT, a novel end-to-end sparse accelerator for Transformers that leverages local similarity to significantly reduce computation and energy consumption while maintaining high accuracy.

Contribution

The paper proposes SPLS, a local similarity-based sparsity prediction mechanism, and architectural innovations enabling efficient end-to-end sparse acceleration of Transformers.

Findings

Reduces total computation by 52.03% with less than 1% accuracy loss.

Achieves 3.29 TOPS/W energy efficiency, outperforming state-of-the-art accelerators.

Improves attention-level energy efficiency by up to 2.95x.

Abstract

Transformers, composed of QKV generation, attention computation, and FFNs, have become the dominant model across various domains due to their outstanding performance. However, their high computational cost hinders efficient hardware deployment. Sparsity offers a promising solution, yet most existing accelerators exploit only intra-row sparsity in attention, while few consider inter-row sparsity. Approaches leveraging inter-row sparsity often rely on costly global similarity estimation, which diminishes the acceleration benefits of sparsity, and typically apply sparsity to only one or two transformer components. Through careful analysis of the attention distribution and computation flow, we observe that local similarity allows end-to-end sparse acceleration with lower computational overhead. Motivated by this observation, we propose ESACT, an end-to-end sparse accelerator for compute-intensive Transformers. ESACT centers on the Sparsity Prediction with Local Similarity (SPLS) mechanism, which leverages HLog quantization to accurately predict local attention sparsity prior to QK generation, achieving efficient sparsity across all transformer components. To support efficient hardware realization, we introduce three architectural innovations. Experimental results on 26 benchmarks demonstrate that SPLS reduces total computation by 52.03% with less than 1% accuracy loss. ESACT achieves an end-to-end energy efficiency of 3.29 TOPS/W, and improves attention-level energy efficiency by 2.95x and 2.26x over SOTA attention accelerators SpAtten and Sanger, respectively.