Low GPU Occupancy Approach to Fast Arithmetic Coding in JPEG2000

Investor logo

Warning

This publication doesn't include Faculty of Education. It includes Faculty of Informatics. Official publication website can be found on muni.cz.
Authors

MATELA Jiří ŠROM Martin HOLUB Petr

Year of publication 2012
Type Article in Proceedings
Conference Lecture Notes in Computer Science 7119
MU Faculty or unit

Faculty of Informatics

Citation
Web SpringerLink
Doi http://dx.doi.org/10.1007/978-3-642-25929-6_13
Field Informatics
Keywords JPEG2000; GPU; arithmetic coding; MQ-Coding; image compression; parallel; low-occupancy; CUDA
Description Arithmetic coding, and especially adaptive MQ-Coder of JPEG2000, is a serial process which does not match well specifics of GPUs as massively parallel processors. In this paper we study and evaluate several approaches to acceleration of MQ-Coding using commodity GPU hardware, including proposal of a new enhanced renormalization procedure. We conclude with "low occupancy approach" and 5.6--16x average speedup when compared to state of the art multi-threaded CPU implementations.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.