arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2301.06672cs.DC
收藏

自定义8位浮点值格式,用于减少近似最近邻搜索中的共享存储体冲突

Custom 8-bit floating point value format for reducing shared memory bank conflict in approximate nearest neighbor search

Hiroyuki Ootomo, Akira Naruse

摘要:

k-最近邻搜索用于各种应用,例如机器学习、计算机视觉、数据库搜索和信息检索。虽然精确最近邻搜索的计算成本巨大,但近似最近邻搜索(ANNS)却引起了人们的广泛关注。IVFPQ是ANNS方法之一。虽然我们可以利用共享内存的高带宽和低延迟来计算NVIDIA GPU上IVFPQ的搜索阶段,但由于共享存储体冲突,吞吐量可能会降低。为了减少存储体冲突和提高搜索吞吐量,我们提出了一种自定义的8位浮点值格式。这种格式没有符号位,可以通过几条指令从FP 32转换为FP 32。我们在GPU上使用这种格式进行IVFPQ,与FP 32和FP 16相比,在没有显著召回损失的情况下实现了更好的性能。

英文摘要:

The k-nearest neighbor search is used in various applications such as machine learning, computer vision, database search, and information retrieval. While the computational cost of the exact nearest neighbor search is enormous, an approximate nearest neighbor search (ANNS) has been attracting much attention. IVFPQ is one of the ANNS methods. Although we can leverage the high bandwidth and low latency of shared memory to compute the search phase of the IVFPQ on NVIDIA GPUs, the throughput can degrade due to shared memory bank conflict. To reduce the bank conflict and improve the search throughput, we propose a custom 8-bit floating point value format. This format doesn't have a sign bit and can be converted from/to FP32 with a few instructions. We use this format for IVFPQ on GPUs and achieved better performance without significant recall loss compared to FP32 and FP16.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑