Convolution of Large 3D Images on GPU and its Decomposition

KARAS, Pavel and David SVOBODA. Convolution of Large 3D Images on GPU and its Decomposition. EURASIP Journal on Advances in Signal Processing. NEW YORK (USA): HINDAWI PUBLISHING CORPORATION, 2011, vol. 2011, No 120, p. 1-12. ISSN 1687-6172. Available from: https://dx.doi.org/10.1186/1687-6180-2011-120.

Basic information

• Original name: Convolution of Large 3D Images on GPU and its Decomposition
• Name in Czech: Konvoluce velkých 3D obrazů na GPU a její dekompozice
• Authors: KARAS, Pavel (203 Czech Republic, guarantor, belonging to the institution) and David SVOBODA (203 Czech Republic, belonging to the institution).
• Edition: EURASIP Journal on Advances in Signal Processing, NEW YORK (USA), HINDAWI PUBLISHING CORPORATION, 2011, 1687-6172.

Other information

• Original language: English
• Type of outcome: Article in a journal
• Field of Study: 10201 Computer sciences, information science, bioinformatics
• Country of publisher: United States of America
• Confidentiality degree: is not subject to a state or trade secret
• WWW: URL
• Impact factor: Impact factor: 1.053 in 2010
• RIV identification code: RIV/00216224:14330/11:00053127
• Organization unit: Faculty of Informatics
• Doi: http://dx.doi.org/10.1186/1687-6180-2011-120
• UT WoS: 000300351600001
• Keywords (in Czech): konvoluce; dekompozice; fft; gpu; dif; 3D
• Keywords in English: convolution; decomposition; fft; gpu; dif; 3D
• Tags: cbia-web
• Tags: International impact, Reviewed

Abstract

In this paper we propose a method for computing convolution of large 3D images. The convolution is performed in a frequency domain using a convolution theorem. The algorithm is accelerated on a graphic card by means of the CUDA parallel computing model. Convolution is decomposed in a frequency domain using the DIF (decimation in frequency) algorithm. We pay attention to keeping our approach efficient in terms of both time and memory consumption and also in terms of memory transfers between CPU and GPU which have a significant influence on overall computational time. We also study the implementation on multiple GPUs and compare the results between the multi-GPU and multi-CPU implementations.

Abstract (in Czech)

Článek se zabývá konvolucí velkých 3D obrazů na grafických kartách a řeší problém pro data, která se nevejdou do paměti GPU.

Links

• LC535, research and development project: Name: Dynamika a organizace chromosomů během buněčného cyklu v normě a patologii

Investor: Ministry of Education, Youth and Sports of the CR, Dynamika a organizace chromosomů během buněčného cyklu v normě a patologii

• MSM0021622419, plan (intention): Name: Vysoce paralelní a distribuované výpočetní systémy

Investor: Ministry of Education, Youth and Sports of the CR, Highly Parallel and Distributed Computing Systems

• MUNI/A/0914/2009, interní kód MU: Name: Rozsáhlé výpočetní systémy: modely, aplikace a verifikace (Acronym: SV-FI MAV)

Investor: Masaryk University, Category A

• 2B06052, research and development project: Name: Vytipování markerů, screening a časná diagnostika nádorových onemocnění pomocí vysoce automatizovaného zpracování multidimenzionálních biomedicínských obrazů (Acronym: Biomarker)

Investor: Ministry of Education, Youth and Sports of the CR, Determination of markers, screening and early diagnostics of cancer diseases using highly automated processing of multidimensional biomedical images