On-the-Fly Lowering Engine: Offloading Data Layout Conversion for Convolutional Neural Networks

Many deep learning frameworks utilize GEneral Matrix Multiplication (GEMM)-based convolution to accelerate CNN execution. GEMM-based convolution provides faster convolution yet requires a data conversion process called lowering (i.e., im2col), which incurs significant memory overhead and diminishes...

Cijeli opis

Bibliografski detalji

Glavni autori:	Mingu Kang, Sangmin Hyun, Tae Hee Han, Jungrae Kim, Seokin Hong
Format:	Artigo
Jezik:	Inglês
Izdano:	IEEE 2022-01-01
Serija:	IEEE Access
Teme:	Convolutional neural network GEMM CPU
Online pristup:	https://ieeexplore.ieee.org/document/9833499/
Oznake:	Dodaj oznaku Bez oznaka, Budi prvi tko označuje ovaj zapis!