RT-HiSS：基于光线追踪加速的高维向量相似性搜索
RT-HiSS: Ray Tracing Accelerated High Dimensional Vector Similarity Searches
浏览论文内容
中文总结 AI 辅助
本研究提出首个基于GPU光线追踪核心的高维向量精确相似性搜索算法RT-HiSS，通过两阶段方法等优化，在六个真实数据集上较现有GPU算法及暴力算法实现显著加速。
中文摘要 AI 辅助
近期的GPU代际产品包含用于图形应用的专用光线追踪（RT）核心。尽管RT核心主要用于渲染，但近期研究表明它们可被用于通用任务，包括相似性搜索。然而，现有方法不支持超过三维的数据集。本研究提出RT-HiSS，这是首个基于GPU RT核心的高维数据集精确相似性搜索算法。GPU相似性搜索在包含大量搜索距离的大型数据集上通常扩展性较差。为解决该问题，RT-HiSS利用RT核心进行快速索引构建与搜索，随后在CUDA核心上进行候选结果优化。我们引入两阶段方法以估计结果规模的上界，在GPU内存约束下实现高效批处理，同时达到近乎完美的负载均衡。此外，我们研究共享内存分块与压缩结果掩码以提升GPU资源利用率。在六个真实世界数据集上，RT-HiSS相较于竞争性的最先进GPU算法可实现最高8.37倍的加速，相较于暴力算法则最高达2368.26倍的加速。
英文摘要
Recent GPU generations include special-purpose ray tracing (RT) cores for graphics applications. While RT cores are primarily used for rendering, recent works show they can be leveraged for general-purpose tasks, including similarity searches. However, existing approaches do not support datasets exceeding three dimensions. In this work, we propose RT-HiSS, the first exact GPU RT-core-based similarity search algorithm for high-dimensional datasets. GPU similarity search often scales poorly for large datasets with substantial search distances. To address this, RT-HiSS uses RT cores for fast index construction and searches, followed by candidate refinement on CUDA cores. We introduce a two-pass approach to estimate an upper bound on result size, enabling efficient batching under GPU memory constraints with near-perfect load balancing. Additionally, we examine shared memory tiling and compressed result masks to improve GPU resource utilization. RT-HiSS yields speedups up to 8.37$\times$ over competitive state-of-the-art GPU algorithms and up to 2,368.26$\times$ relative to the brute-force algorithm across six real-world datasets.
发表机构
- Northern Arizona University（北亚利桑那大学）
机构由 AI 辅助整理，请以论文原文为准。
