High Performance and Portable Convolution Operators for ARM-based Multicore Processors

TL;DR

This paper introduces a portable, high-performance convolution algorithm for ARM multicore processors that eliminates intermediate memory and IM2COL transform costs by leveraging the BLIS GEMM kernel structure.

Contribution

It presents a novel convolution method that avoids intermediate memory and reduces computation time, enhancing performance and portability on ARM multicore systems.

Findings

Achieves high performance comparable to existing methods

Reduces memory usage significantly

Maintains portability across ARM architectures

Abstract

The considerable impact of Convolutional Neural Networks on many Artificial Intelligence tasks has led to the development of various high performance algorithms for the convolution operator present in this type of networks. One of these approaches leverages the \imcol transform followed by a general matrix multiplication (GEMM) in order to take advantage of the highly optimized realizations of the GEMM kernel in many linear algebra libraries. The main problems of this approach are 1) the large memory workspace required to host the intermediate matrices generated by the IM2COL transform; and 2) the time to perform the IM2COL transform, which is not negligible for complex neural networks. This paper presents a portable high performance convolution algorithm based on the BLIS realization of the GEMM kernel that avoids the use of the intermediate memory by taking advantage of the BLIS structure. In addition, the proposed algorithm eliminates the cost of the explicit IM2COL transform, while maintaining the portability and performance of the underlying realization of GEMM in BLIS.