Bit-GraphBLAS: Bit-Level Optimizations of Matrix-Centric Graph Processing on GPU

TL;DR

This paper introduces bit-level graph representations and GPU optimizations, significantly accelerating matrix-centric graph algorithms like BFS, SSSP, PR, CC, and TC.

Contribution

It proposes the B2SR bit-level graph representation and GPU-specific optimizations, filling a gap in existing matrix-centric graph processing frameworks.

Findings

Up to 40x speedup for SpMV kernel

Up to 6555x speedup for SpGEMM kernel

Graph algorithms like BFS and SSSP are significantly accelerated

Abstract

In a general graph data structure like an adjacency matrix, when edges are homogeneous, the connectivity of two nodes can be sufficiently represented using a single bit. This insight has, however, not yet been adequately exploited by the existing matrix-centric graph processing frameworks. This work fills the void by systematically exploring the bit-level representation of graphs and the corresponding optimizations to the graph operations. It proposes a two-level representation named Bit-Block Compressed Sparse Row (B2SR) and presents a series of optimizations to the graph operations on B2SR by leveraging the intrinsics of modern GPUs. Evaluations on NVIDIA Pascal and Volta GPUs show that the optimizations bring up to $40\times$ and $6555\times$ for essential GraphBLAS kernels SpMV and SpGEMM, respectively, making GraphBLAS-based BFS accelerate up to $433\times$, SSSP, PR, and CC up to $35\times$, and TC up to $52\times$.