A complete discussion on fully reconfigurable, digital, scalable, graph and sparsity-aware near-memory accelerator for graph neural networks

TL;DR

This paper introduces NEM-GNN, a scalable, energy-efficient, reconfigurable near-memory accelerator for GNNs that handles heterogeneous memory behaviors effectively.

Contribution

It proposes a novel PIM architecture with early termination, pre-computation, and sparsity-aware aggregation, overcoming scalability and energy limitations of prior approaches.

Findings

Achieves 80-230x higher performance than prior methods.

Demonstrates 80-300x higher throughput.

Provides 850-1134x better energy efficiency.

Abstract

Graph neural networks (GNNs) have gained significant interest for applications such as citation network analysis and drug discovery due to their ability to apply machine learning techniques on graph-structured data. GNNs typically employ a two-stage execution pipeline consisting of combination and aggregation kernels. The combination stage performs data-intensive convolution operations with relatively regular memory access patterns, whereas the aggregation stage operates on sparse graph data with highly irregular accesses. These heterogeneous memory behaviors make conventional CPU- and GPU-based execution energy inefficient due to substantial data movement overheads. Existing accelerators attempt to mitigate these challenges using specialized architectures and processing-in-memory (PIM) techniques. However, prior approaches often suffer from scalability limitations, area overheads, restricted parallelism, and energy inefficiencies associated with analog compute and dedicated accelerator structures. This paper presents NEM-GNN, a scalable DAC/ADC-less processing-in-memory architecture for graph neural network acceleration. The proposed design introduces early compute termination mechanisms, pre-computation using reconfigurable system-on-chip components, and graph- and sparsity-aware near-memory aggregation using a compute-as-soon-as-ready (CAR) and broadcast-based execution model. Experimental results demonstrate that NEM-GNN achieves approximately 80--230x higher performance, 80--300x higher throughput, 850--1134x better energy efficiency, and 7--8x higher compute density compared to prior state-of-the-art approaches.