Similarity Search with Tensor Core Units
This work addresses efficiency in similarity search for applications using TCU hardware, but it is incremental as it adapts existing algorithms to new hardware.
The paper tackles the problem of accelerating similarity search by leveraging Tensor Core Units (TCUs), achieving a √m speedup in algorithms for Johnson-Lindenstrauss dimensionality reduction and similarity join compared to traditional methods.
Tensor Core Units (TCUs) are hardware accelerators developed for deep neural networks, which efficiently support the multiplication of two dense $\sqrt{m}\times \sqrt{m}$ matrices, where $m$ is a given hardware parameter. In this paper, we show that TCUs can speed up similarity search problems as well. We propose algorithms for the Johnson-Lindenstrauss dimensionality reduction and for similarity join that, by leveraging TCUs, achieve a $\sqrt{m}$ speedup up with respect to traditional approaches.