High-performance Video Decoding using Graphics Processing Units | Dissertationsschrift
Biao Wang
ISBN 10: 3746731003 ISBN 13: 9783746731001
Verlag: epubli, 2018
Neu
Taschenbuch
Verkäufer
preigu, Osnabrück, Deutschland
Verkäuferbewertung 5 von 5 Sternen
AbeBooks-Verkäufer seit 5. August 2024
Dieser Artikel ist nicht mehr verfügbar.
Beschreibung
Beschreibung:
High-performance Video Decoding using Graphics Processing Units | Dissertationsschrift | Biao Wang | Taschenbuch | Englisch | epubli | EAN 9783746731001 | Verantwortliche Person für die EU: preigu GmbH & Co. KG, Lengericher Landstr. 19, 49078 Osnabrück, mail[at]preigu[dot]de | Anbieter: preigu. Bestandsnummer des Verkäufers 118672227
Inhaltsangabe:
The increasing demand of decoding high-quality videos can lead to a challenging com- putational requirement for conventional Central Processing Unit (CPU) architectures. Graphics Processing Units (GPUs) in general provide higher computational power than CPUs. Efficient GPU execution, however, requires massive parallelism and little ex- ecuting divergence, two criteria are not fully met by all video decoding kernels. This thesis exploits how GPUs can be effectively used in video decoding applications. The challenges include proper workload distribution between the CPU and GPU, task optimizations on two heterogeneous devices, and efficient communication between them. A complete parallel HEVC decoder was proposed for heterogeneous CPU+GPU systems. We exploited available decoding parallelism on the CPU, GPU, and between the two devices simultaneously. On top of the parallel design, two workload balancing schemes were implemented, in order to adapt computation resource variation on CPU and GPU. In addition, an energy measurement module was developed for energy efficiency analysis. Evaluated results showed that suitable decoding kernels can be accelerated substan- tially (up to 28.2×) on GPUs at the kernel level. At the application level, using GPU architecture can provide significant acceleration when only a low number (1 to 8) of CPU cores are available. On a system consisting of an NVIDIA Titan X Maxwell GPU and an Intel Xeon E5-2699v3 CPU, with four CPU cores, the proposed HEVC decoder delivers 167 frames per second for 4K videos, corresponding to a speedup of 2.2× over the state- of-the-art CPU decoder using four CPU cores. When more CPU cores (>8) are employed, the benefit of using GPU vanishes and the performance is eventually outperformed by the CPU decoder due to GPU overloading. With respect to energy, because of its high power consumption GPU architecture is not as efficient as the CPU for HEVC decoding.
Über die Autorin bzw. den Autor: Biao Wang received the M.S. degree in Computer Application Technology in July 2010 and Bachelor degree in Computer Software Technology in July 2007, both from University of Electronic Science and Technology of China (UESTC), ChengDu, China. Since Oct 2010, he joined the Embedded Systems Architecture as a PhD student with scholarship from China Scholarship Council (CSC). His research is focused on video decoding using Graphic Processing Units (GPUs). He is experienced in high performance video decoding on heterogeneous systems with CPU and GPU. Some of these works are published in international conferences and prestigious journals.
„Über diesen Titel“ kann sich auf eine andere Ausgabe dieses Titels beziehen.
Bibliografische Details
Titel: High-performance Video Decoding using ...
Verlag: epubli
Erscheinungsdatum: 2018
Einband: Taschenbuch
Zustand: Neu
Beste Suchergebnisse beim ZVAB
Foto des Verkäufers
High-performance Video Decoding using Graphics Processing Units : Dissertationsschrift
Anbieter: AHA-BUCH GmbH, Einbeck, Deutschland
Verkäuferbewertung 5 von 5 Sternen
Taschenbuch. Zustand: Neu. Druck auf Anfrage Neuware - Printed after ordering - The increasing demand of decoding high-quality videos can lead to a challenging com-putational requirement for conventional Central Processing Unit (CPU) architectures.Graphics Processing Units (GPUs) in general provide higher computational powerthan CPUs. Efficient GPU execution, however, requires massive parallelism and little ex-ecuting divergence, two criteria are not fully met by all video decoding kernels. This thesisexploits how GPUs can be effectively used in video decoding applications. The challengesinclude proper workload distribution between the CPU and GPU, task optimizations ontwo heterogeneous devices, and efficient communication between them.A complete parallel HEVC decoder was proposed for heterogeneous CPU+GPUsystems. We exploited available decoding parallelism on the CPU, GPU, and between thetwo devices simultaneously. On top of the parallel design, two workload balancing schemeswere implemented, in order to adapt computation resource variation on CPU and GPU.In addition, an energy measurement module was developed for energy efficiency analysis.Evaluated results showed that suitable decoding kernels can be accelerated substan-tially (up to 28.2×) on GPUs at the kernel level. At the application level, using GPUarchitecture can provide significant acceleration when only a low number (1 to 8) of CPUcores are available. On a system consisting of an NVIDIA Titan X Maxwell GPU and anIntel Xeon E5-2699v3 CPU, with four CPU cores, the proposed HEVC decoder delivers167 frames per second for 4K videos, corresponding to a speedup of 2.2× over the state-of-the-art CPU decoder using four CPU cores. When more CPU cores (>8) are employed,the benefit of using GPU vanishes and the performance is eventually outperformed by theCPU decoder due to GPU overloading. With respect to energy, because of its high powerconsumption GPU architecture is not as efficient as the CPU for HEVC decoding. Artikel-Nr. 9783746731001
Neu kaufen
EUR 46,58
EUR 62,37 shipping
Versand von Deutschland nach USA
Versand von Deutschland nach USA
Anzahl: 2 verfügbar