Over the last decade, significant research effort has been put into improving the performance of hash join operation on GPUs. Over the same period, there have been significant changes to the GPU architecture. Hence in this paper, we first revisit the major GPU hash join implementations in the last decade and detail how they take advantage of different GPU architecture features. We then perform a comprehensive performance evaluation of these implementations using different generations of GPUs released over the last decade, which helps to shed light on the impact of different architecture features and to identify the factors guiding the choice of these features. We then study how data characteristics like skew and match rate impact the performance of GPU hash join implementations and propose techniques to improve the performance of existing implementations under such conditions. Finally, we perform an in-depth comparison of the performance and cost-efficiency of GPU hash join implementations against state-of-the-art CPU implementation.

中文翻译：

---

重新审视图形处理器上的哈希连接：十年后

在过去的十年中，大量的研究工作已经投入到提高 GPU 上散列连接操作的性能上。在同一时期，GPU 架构发生了重大变化。因此，在本文中，我们首先回顾了过去十年中主要的 GPU 哈希连接实现，并详细说明它们如何利用不同的 GPU 架构特性。然后，我们使用过去十年发布的不同代 GPU 对这些实现进行综合性能评估，这有助于阐明不同架构特征的影响，并确定指导这些特征选择的因素。然后，我们研究偏斜和匹配率等数据特征如何影响 GPU 哈希连接实现的性能，并提出在这种情况下提高现有实现性能的技术。最后，我们对 GPU 哈希连接实现与最先进的 CPU 实现的性能和成本效率进行了深入比较。