An Efficient Sparse Inference Software Accelerator for Transformer-based Language Models on CPUs

TL;DR

This paper introduces a highly efficient sparse inference software accelerator for Transformer-based language models on CPUs, leveraging structured pruning and Intel hardware features to significantly improve inference speed and support industrial latency requirements.

Contribution

The paper presents a novel sparse deep learning inference stack optimized for CPUs, with a new SpMM kernel that outperforms existing libraries and accelerates Transformer models under real-world constraints.

Findings

SpMM kernel outperforms existing sparse libraries by an order of magnitude.

Achieves up to 5x speedup over dense GEMM in industry-standard libraries.

Demonstrates up to 1.5x speedup over neural network inference on real models.

Abstract

In recent years, Transformer-based language models have become the standard approach for natural language processing tasks. However, stringent throughput and latency requirements in industrial applications are limiting their adoption. To mitigate the gap, model compression techniques such as structured pruning are being used to improve inference efficiency. However, most existing neural network inference runtimes lack adequate support for structured sparsity. In this paper, we propose an efficient sparse deep learning inference software stack for Transformer-based language models where the weights are pruned with constant block size. Our sparse software accelerator leverages Intel Deep Learning Boost to maximize the performance of sparse matrix - dense matrix multiplication (commonly abbreviated as SpMM) on CPUs. Our SpMM kernel outperforms the existing sparse libraries (oneMKL, TVM, and LIBXSMM) by an order of magnitude on a wide range of GEMM shapes under 5 representative sparsity ratios (70%, 75%, 80%, 85%, 90%). Moreover, our SpMM kernel shows up to 5x speedup over dense GEMM kernel of oneDNN, a well-optimized dense library widely used in industry. We apply our sparse accelerator on widely-used Transformer-based language models including Bert-Mini, DistilBERT, Bert-Base, and BERT-Large. Our sparse inference software shows up to 1.5x speedup over Neural Magic's Deepsparse under same configurations on Xeon on Amazon Web Services under proxy production latency constraints. We also compare our solution with two framework-based inference solutions, ONNX Runtime and PyTorch, and demonstrate up to 37x speedup over ONNX Runtime and 345x over PyTorch on Xeon under the latency constraints. All the source code is publicly available on Github: https://github.com/intel/intel-extension-for-transformers.