RAPID-Graph: Recursive All-Pairs Shortest Paths Using Processing-in-Memory for Dynamic Programming on Graphs

TL;DR

RAPID-Graph introduces a processing-in-memory system with novel algorithms and architecture optimizations to significantly accelerate all-pairs shortest paths computations on large graphs, achieving substantial speed and energy efficiency improvements.

Contribution

It presents a co-designed PIM system with a recursion-aware partitioner and a 2.5D memory stack, enabling exact APSP computation efficiently within digital PIM arrays.

Findings

5.8x faster than GPU clusters on large datasets

1186x more energy efficient than state-of-the-art GPU solutions

Up to 42.8x speedup over NVIDIA H100 GPU

Abstract

All-pairs shortest paths (APSP) remains a major bottleneck for large-scale graph analytics, as data movement with cubic complexity overwhelms the bandwidth of conventional memory hierarchies. In this work, we propose RAPID-Graph to address this challenge through a co-designed processing-in-memory (PIM) system that integrates algorithm, architecture, and device-level optimizations. At the algorithm level, we introduce a recursion-aware partitioner that enables an exact APSP computation by decomposing graphs into vertex tiles to reduce data dependency, such that both Floyd-Warshall and Min-Plus kernels execute fully in-place within digital PIM arrays. At the architecture and device levels, we design a 2.5D PIM stack integrating two phase-change memory compute dies, a logic die, and high-bandwidth scratchpad memory within a unified advanced package. An external non-volatile storage stack stores large APSP results persistently. The design achieves both tile-level and unit-level parallel processing to sustain high throughput. On the 2.45M-node OGBN-Products dataset, RAPID-Graph is 5.8x faster and 1,186x more energy efficient than state-of-the-art GPU clusters, while exceeding prior PIM accelerators by 8.3x in speed and 104x in efficiency. It further delivers up to 42.8x speedup and 392x energy savings over an NVIDIA H100 GPU.