Efficient Convolutional Dataflows on Low-Power Neural Network Accelerators


METADATA ONLY
Loading...

Date

2024-09

Publication Type

Journal Article

ETH Bibliography

yes

Citations

Altmetric
METADATA ONLY

Data

Rights / License

Abstract

Dilated and transposed convolutions are widely used in modern convolutional neural networks (CNNs). These kernels are used extensively during CNN training and inference of applications such as image segmentation and high-resolution image generation. We find that commonly-used low-power CNN inference accelerators are not optimized for both these convolutional kernels. Dilated and transposed convolutions introduce significant zero padding when mapped to the underlying spatial architecture, significantly degrading performance and energy efficiency. Existing approaches that address this issue require significant design changes to the otherwise simple, efficient, and well-adopted architectures used to compute direct convolutions. To address this challenge, we propose EcoFlow, a new set of dataflows and mapping algorithms for dilated and transposed convolutions. These algorithms are tailored to execute efficiently on existing low-cost, small-scale spatial architectures and requires minimal changes to existing accelerators. At its core, EcoFlow eliminates zero padding through careful dataflow orchestration and data mapping tailored to the spatial architecture. We evaluate EcoFlow on CNN training workloads and Generative Adversarial Network (GAN) workloads. Experiments in our new cycle-accurate simulator show that, using a common CNN inference accelerator, EcoFlow 1) reduces end-to-end CNN training time between 7-85%, and 2) improves end-to-end GAN training performance between 29-42%, compared to state-of-the-art CNN dataflows.

Publication status

published

Editor

Book title

Volume

73 (9)

Pages / Article No.

2275 - 2289

Publisher

IEEE

Event

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Convolutional neural network; Hardware accelerators

Organisational unit

09483 - Mutlu, Onur / Mutlu, Onur check_circle

Notes

Funding

Related publications and datasets

Is new version of: