Ansor: Generating High-Performance Tensor Programs for Deep Learning

TL;DR

Ansor is a framework that automatically generates high-performance tensor programs for deep learning across various hardware platforms by exploring a vast search space and fine-tuning programs with advanced strategies.

Contribution

Ansor introduces a hierarchical search and learned cost model to discover tensor programs beyond existing methods, improving deep learning performance on multiple hardware types.

Findings

Achieves up to 3.8x speedup on Intel CPU

Improves performance by up to 2.6x on ARM CPU

Enhances GPU deep learning execution by 1.7x

Abstract

High-performance tensor programs are crucial to guarantee efficient execution of deep neural networks. However, obtaining performant tensor programs for different operators on various hardware platforms is notoriously challenging. Currently, deep learning systems rely on vendor-provided kernel libraries or various search strategies to get performant tensor programs. These approaches either require significant engineering effort to develop platform-specific optimization code or fall short of finding high-performance programs due to restricted search space and ineffective exploration strategy. We present Ansor, a tensor program generation framework for deep learning applications. Compared with existing search strategies, Ansor explores many more optimization combinations by sampling programs from a hierarchical representation of the search space. Ansor then fine-tunes the sampled programs with evolutionary search and a learned cost model to identify the best programs. Ansor can find high-performance programs that are outside the search space of existing state-of-the-art approaches. In addition, Ansor utilizes a task scheduler to simultaneously optimize multiple subgraphs in deep neural networks. We show that Ansor improves the execution performance of deep neural networks relative to the state-of-the-art on the Intel CPU, ARM CPU, and NVIDIA GPU by up to $3.8\times$, $2.6\times$, and $1.7\times$, respectively.