SAGE: A Storage-Based Approach for Scalable and Efficient Sparse Generalized Matrix-Matrix Multiplication

TL;DR

SAGE introduces a storage-based approach for scalable, efficient sparse matrix multiplication that leverages SSDs and a three-layer architecture to outperform existing methods in large-scale network analysis.

Contribution

The paper proposes SAGE, a novel storage-based SpGEMM method that reduces memory bottlenecks and communication overhead, enabling scalable and efficient large-scale network processing.

Findings

SAGE outperforms existing methods in scalability and efficiency.

SAGE effectively balances workloads and reduces I/O overhead.

The approach prevents buffer overflows with distribution-aware memory allocation.

Abstract

Sparse generalized matrix-matrix multiplication (SpGEMM) is a fundamental operation for real-world network analysis. With the increasing size of real-world networks, the single-machine-based SpGEMM approach cannot perform SpGEMM on large-scale networks, exceeding the size of main memory (i.e., not scalable). Although the distributed-system-based approach could handle large-scale SpGEMM based on multiple machines, it suffers from severe inter-machine communication overhead to aggregate results of multiple machines (i.e., not efficient). To address this dilemma, in this paper, we propose a novel storage-based SpGEMM approach (SAGE) that stores given networks in storage (e.g., SSD) and loads only the necessary parts of the networks into main memory when they are required for processing via a 3-layer architecture. Furthermore, we point out three challenges that could degrade the overall performance of SAGE and propose three effective strategies to address them: (1) block-based workload allocation for balancing workloads across threads, (2) in-memory partial aggregation for reducing the amount of unnecessarily generated storage-memory I/Os, and (3) distribution-aware memory allocation for preventing unexpected buffer overflows in main memory. Via extensive evaluation, we verify the superiority of SAGE over existing SpGEMM methods in terms of scalability and efficiency.