Understanding and bridging the gaps in current GNN performance optimizations
Understanding and bridging the gaps in current GNN performance optimizations
复制标题
DOI：
10.1145/3437801.3441585
复制
发表时间：
2021-02
期刊：
影响因子：
--
通讯作者：
Kezhao Huang;Jidong Zhai;Zhen Zheng;Youngmin Yi;Xipeng Shen
中科院分区：
文献类型：
--
作者：
Kezhao Huang;Jidong Zhai;Zhen Zheng;Youngmin Yi;Xipeng Shen
图神经网络（GNN）最近在许多领域引起了快速增长的兴趣，因为它在图上学习的有效性。最大限度地提高其性能对于许多任务至关重要，但仍然是初步了解。在这项工作中，我们深入研究了最先进的GNN框架，揭示了当前框架在优化GNN性能方面的五个主要差距，特别是在处理GNN相对于传统图或DNN操作的特殊复杂性方面。基于这些见解，我们提出了一系列优化来填补空白。这些优化利用了最先进的GPU优化技术，并根据GNN的特殊属性对其进行了调整。实验结果表明，在不同的GNN模型上，这些优化方法的性能比现有框架提高了1.37×-15.5×.
Graph Neural Network (GNN) has recently drawn a rapid increase of interest in many domains for its effectiveness in learning over graphs. Maximizing its performance is essential for many tasks, but remains preliminarily understood. In this work, we provide an in-depth examination of the state-of-the-art GNN frameworks, revealing five major gaps in the current frameworks in optimizing GNN performance, especially in handling the special complexities of GNN over traditional graph or DNN operations. Based on the insights, we put together a set of optimizations to fill the gaps. These optimizations leverage the state-of-the-art GPU optimization techniques and tailor them to the special properties of GNN. Experimental results show that these optimizations achieve 1.37×--15.5× performance improvement over the state-of-the-art frameworks on various GNN models.