GPU-acceleration for Large-scale Tree Boosting

TL;DR

This paper introduces a GPU-accelerated, histogram-based algorithm for decision tree building that significantly improves the scalability and speed of training gradient boosting models without sacrificing accuracy.

Contribution

It develops a novel GPU-efficient histogram construction method for decision trees, enabling faster training in boosting systems compared to existing CPU and GPU algorithms.

Findings

7-8 times faster than CPU-based histogram algorithms in LightGBM

25 times faster than exact-split algorithms in XGBoost

Achieves similar accuracy with significantly reduced training time

Abstract

In this paper, we present a novel massively parallel algorithm for accelerating the decision tree building procedure on GPUs (Graphics Processing Units), which is a crucial step in Gradient Boosted Decision Tree (GBDT) and random forests training. Previous GPU based tree building algorithms are based on parallel multi-scan or radix sort to find the exact tree split, and thus suffer from scalability and performance issues. We show that using a histogram based algorithm to approximately find the best split is more efficient and scalable on GPU. By identifying the difference between classical GPU-based image histogram construction and the feature histogram construction in decision tree training, we develop a fast feature histogram building kernel on GPU with carefully designed computational and memory access sequence to reduce atomic update conflict and maximize GPU utilization. Our algorithm can be used as a drop-in replacement for histogram construction in popular tree boosting systems to improve their scalability. As an example, to train GBDT on epsilon dataset, our method using a main-stream GPU is 7-8 times faster than histogram based algorithm on CPU in LightGBM and 25 times faster than the exact-split finding algorithm in XGBoost on a dual-socket 28-core Xeon server, while achieving similar prediction accuracy.