Accelerating Barnes-Hut t-SNE Algorithm by Efficient Parallelization on Multi-Core CPUs

TL;DR

This paper presents Acc-t-SNE, a highly optimized CPU implementation of Barnes-Hut t-SNE that significantly accelerates high-dimensional data visualization tasks through advanced parallelization and cache optimization techniques.

Contribution

The paper introduces a new CPU-based Barnes-Hut t-SNE implementation with novel optimizations, achieving substantial speedups over existing methods.

Findings

Up to 261x faster than scikit-learn

Up to 4x faster than daal4py implementation

Effective parallelization on 32-core CPUs

Abstract

t-SNE remains one of the most popular embedding techniques for visualizing high-dimensional data. Most standard packages of t-SNE, such as scikit-learn, use the Barnes-Hut t-SNE (BH t-SNE) algorithm for large datasets. However, existing CPU implementations of this algorithm are inefficient. In this work, we accelerate the BH t-SNE on CPUs via cache optimizations, SIMD, parallelizing sequential steps, and improving parallelization of multithreaded steps. Our implementation (Acc-t-SNE) is up to 261x and 4x faster than scikit-learn and the state-of-the-art BH t-SNE implementation from daal4py, respectively, on a 32-core Intel(R) Icelake cloud instance.