Work

Optimizing General Sparse Matrix-Matrix Multiplication on the GPU

Year: 2025

Type: article

Abstract: General Sparse Matrix-Matrix Multiplication (SpGEMM) is a crucial computational kernel in the field of scientific and engineering computing. Due to the irregular distribution of nonzero elements in sparse matrices, SpGEMM computation faces challenges such as non-contiguous memory access and workload imbalance. This article focuses on optimizing SpGEMM for GPU platforms. First, a lightweight machine learning model is trained to predict the optimal method for estimating the size of result matrix. Next, different kernels are launched in groups to maximize GPU shared memory utilization and achieve load balancing. For the hash-based sparse accumulator, heuristic methods are used to select the optimal hash load factors and hash multiplier factors, thereby reducing the number of hash collisions. In addition, thread reduction is applied in the symbolic phase to enhance intra-block parallelism. Combining these optimization strategies, we implemented an adaptive SpGEMM algorithm for GPUs and compared its performance with current state-of-the-art algorithms. The results show that our algorithm achieves significant performance improvements.

Source: ACM Transactions on Architecture and Code Optimization

DOI: 10.1145/3774654

Authors: Yizhuo Wang, Hongpeng Lin, Bingxin Wei, Jianhua Gao, Weixing Ji

Institutions: Beijing Institute of Technology, Beijing Normal University

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

Funder: National Science and Technology Major Project

Award: G7573827884