Understanding sources of inefficiency in general-purpose chips
Published 19 June 2010
Rehan Hameed, Wajahat Qadeer, Megan Wachs, Omid Azizi, Alex Solomatnikov, Benjamin C. Lee
Citations393
Generate an AI Snapshot to get a quick, structured summary of this paper.
Study Snapshot
ObjectiveStudy objective
MethodsResearch methodology
PopulationPopulation studied
Sample sizeSample sizes
OutcomesStudy outcomes here
ResultsStudy results comes here
LimitationsResearch study limitations comes here
A concise AI-generated summary of the paper will appear here once you click Generate AI Snapshot.
TL;DR
The sources of performance and energy overheads in general-purpose processing systems are explored by quantifying the overheads of a 720p HD H.264 encoder running on a general-purpose CMP system and methods to eliminate these overheads are explored by transforming the CPU into a specialized system for H.264 encoding.
Abstract
Due to their high volume, general-purpose processors, and now chip multiprocessors (CMPs), are much more cost effective than ASICs, but lag significantly in terms of performance and energy efficiency. This paper explores the sources of these performance and energy overheads in general-purpose processing systems by quantifying the overheads of a 720p HD H.264 encoder running on a general-purpose CMP system. It then explores methods to eliminate these overheads by transforming the CPU into a specialized system for H.264 encoding. We evaluate the gains from customizations useful to broad classes of algorithms, such as SIMD units, as well as those specific to particular computation, such as customized storage and functional units.
Keywords
Computer ScienceEngineering
IEEE Transactions on Circuits and Systems for Video TechnologyOverview of the H.264/AVC video coding standard
8,022 Citations2003Thomas Wiegand, Gary J. Sullivan +2 more
An overview of the technical features of H.264/AVC is provided, profiles and applications for the standard are described, and the history of the standardization process is outlined.
IEEE Journal of Solid-State CircuitsDesign of ion-implanted MOSFET's with very small physical dimensions
3,477 Citations1974R.H. Dennard, F.H. Gaensslen +4 more
This paper considers the design, fabrication, and characterization of very small Mosfet switching devices suitable for digital integrated circuits, using dimensions of the order of 1 /spl mu/.
IEEE Transactions on Circuits and Systems for Video TechnologyAnalysis and architecture design of an HDTV720p 30 frames/s H.264/AVC encoder
282 Citations2006Tung-Chien Chen, Shao‐Yi Chien +5 more
The four-stage macroblock pipelined system architecture is proposed with an efficient scheduling and memory hierarchy, and the prototype chip of the efficient H.264/AVC video encoder for HDTV applications is implemented.
Scaling, power, and the future of CMOS
279 Citations2006Mark Horowitz, E. Alon +4 more
The forces that caused the power problem, the solutions that were applied, and what the solutions tell us about the problem are reviewed, and possible strategies for dealing with the slowdown in gate energy scaling are described.
IEEE Transactions on Circuits and Systems I Fundamental Theory and ApplicationsAnalysis and architecture design of variable block-size motion estimation for H.264/AVC
245 Citations2006Ching-Yeh Chen, Shao‐Yi Chien +4 more
Two hardware architectures are proposed that can support traditional fixed block-size motion estimation as well as VBSME with less chip area overhead compared to previous approaches and an eight-parallel SAD tree with a shared reference buffer for H.264/AVC integer motion estimation is proposed.
Application-specific instruction generation for configurable processor architectures
236 Citations2004Jason Cong, Yiping Fan +2 more
A set of algorithms, including pattern generation, pattern selection, and application mapping, are proposed to efficiently utilize the instruction set extensibility of the target configurable processor.
Fast mode decision and motion estimation for JVT/H.264
193 Citations2004Peng Yin, H.-Y.C. Tourapis +2 more
A new scheme to jointly optimize motion estimation and mode decision is proposed and results show that up to 90% complexity reduction while maintaining coding efficiency.
IEEE Transactions on ComputersAutomated Custom Instruction Generation for Domain-Specific Processor Acceleration
117 Citations2005Nathan Clark, Hongtao Zhong +1 more
The design of a system to automate the instruction set customization process is presented and generalization techniques are presented which enable the application-specific hardware to be more effectively used across a domain.
Fully utilized and reusable architecture for fractional motion estimation of H.264/AVC
116 Citations2004Tung-Chien Chen, Yu‐Wen Huang +1 more
A new VLSI architecture for fractional motion estimation of the H.264/AVC video compression standard is contributed, characterized by a reusable feature, that can support situations in different specifications, multiple standards, fast algorithms and some cost considerations.
IEEE Journal of Solid-State CircuitsA design environment for high-throughput low-power dedicated signal processing systems
112 Citations2002W. Rhett Davis, N. Zhang +8 more
Customizable Embedded Processors: Design Technologies and Applications
107 Citations2006Paolo Ienne, Rainer Leupers
This book is the first to explore comprehensively one of the most fundamental trends which emerged in the last decade: to treat processors not as rigid, fixed entities, which designers include "as is" in their products; but rather, to build sound methodologies to tailor-fit processors to the specific needs of such products.
Journal of Visual Communication and Image RepresentationImplementation of H.264 encoder and decoder on personal computers
88 Citations2005Yen-Kuang Chen, Eric Q. Li +2 more
It is demonstrated that hardware-specific algorithm modifications can speed up the H.264 decoder and encoder substantially, and the performance improvement techniques on modern microprocessors demonstrated in this work can be applied not only to H.265, but also to other video or multimedia processing applications.
IEEE Transactions on Circuits and Systems for Video TechnologyHigh-Throughput Architecture for H.264/AVC CABAC Compression System
82 Citations2006Roberto R. Osorio, J.D. Bruguera
A fast and new architecture for arithmetic coding adapted to the characteristics of CABAC, including optimized use of memory and context managing and fast processing able to encode more than two symbols per cycle is presented.
AnySP
81 Citations2009Mark Woh, Sangwon Seo +4 more
AnySP, a fully programmable architecture that targets multiple application domains, addresses the challenges for next-generation mobile signal processing.
A 242mW, 10mm <sup>2</sup> 1080p H.264/AVC high profile encoder chip
59 Citations2008Yu-Kun Lin, Dewei Li +6 more
A 1080 p high profile H.264 encoder is designed by the robust reusable silicon IP methodology and fabricated in a 0.13 mum CMOS technology with an area of 10 mm2 and 242 mW at 145 MHz.
IEEE Computer Architecture LettersAn Energy-Efficient Processor Architecture for Embedded Systems
55 Citations2008James Balfour, William J. Dally +3 more
An efficient programmable architecture for compute-intensive embedded applications that uses instruction registers to reduce the cost of delivering instructions, and a hierarchical and distributed data register organization to deliver data.
A 7mW-to-183mW Dynamic Quality-Scalable H.264 Video Encoder Chip
51 Citations2007Hsiu-Cheng Chang, Jiawei Chen +10 more
A dynamic quality-scalable H.264 video encoder is presented for power-adaptive video encoding and, compared to the state-of-the-art design for real-time HD720 video encoding, a 49% reduction in gate count and a 61% reduced in internal memory are achieved.
Characteristics of workloads used in high performance and technical computing
42 Citations2007Razvan Cheveresan, Matt Ramsay +2 more
The results of this work show that the HPC application space is surprisingly diverse, with some codes showing similar data sharing and locality properties with commercial applications.
Real-time H.264/AVC codec on intel architectures
27 Citations2005V. Iverson, Jeff McVeigh +1 more
These results demonstrate that high-quality, real- time H.264/AVC recording and playback are achievable on today's PC platforms and that real-time playback on handheld platforms can also be achieved, with both implemented as optimized software for these general-purpose processor architectures.
Scaling, Power and the Future of CMOS
26 Citations2007Mark Horowitz
This talk will review power optimized design methods and shows how power is strongly tied to performance and that variability aversely effects power efficiency, and shows that unless die size shrinks, in future technologies most of the devices will need to be idle most ofThe time which has strong ramifications for the both the underlying device and system design.
A memory system design framework
22 Citations2009Amin Firoozshahian, Alex Solomatnikov +5 more
A microarchitecture framework for cache and protocol controllers, which can aid in generating the RTL for new memory systems, and decomposes the memory model into sequences of state and data manipulations, which produces a polymorphic memory system.
A VLSI architecture design of an edge based fast intra prediction mode decision algorithm for h.264/avc
18 Citations2007Shen Li, Xianghui Wei +2 more
A novel VLSI design is presented of an edge based fast intra MD algorithm which can constantly reduce about 66% of the RD related computation with a negligible quality loss and is expected to be utilized as a favorable accelerator hardware module in a real-time HDTV (1920-1088p) H.264 encoder or MPEG2-H.264 transcoder.
Scientific applications vs. SPEC-FP
14 Citations2006Kyle Rupnow, Arun Rodrigues +2 more
This work examines common dataflow and usage patterns of integer instructions---information essential to develop hardware techniques to accelerate critical scientific applications, and presents statistics for SPEC-FP and Sandia applications.
Proceedings - ACM IEEE Design Automation ConferenceChip multi-processor generator
14 Citations2007Alex Solomatnikov, Amin Firoozshahian +7 more
This paper argues that flexible, general computing chips are trying to accomplish more than is commercially needed, and proposes a two-step design system that combines the program and configuration, tailoring the original framework to create a chip that is optimized toward the desired set of applications.
Flexible architectures for engineering successful SOCs
13 Citations2004Chris Rowen, Steve Leibson
This paper focuses on a particular SOC design technology and methodology, here called the advanced or processor-centric SOC design method, which reduces the risk of SOC design and increases ROI hy using configurahle processors to implement onchip functions while increasing the SOC's flexibility through software programmability.
Using a configurable processor generator for computer architecture prototyping
11 Citations2009Alex Solomatnikov, Amin Firoozshahian +6 more
Having a sophisticated software tool chain not only allowed us to run more applications on the authors' machine, it once again pointed out the need to use optimized code to get an accurate evaluation of architectural features.
Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIEA VLSI architecture for high performance CABAC encoding
9 Citations2005Hassan Shojania, Subramania I. Sudharsanan
This paper provides an efficient, pipelined VLSI architecture for CABAC encoding along with an analysis of critical issues and an FPGA implementation of the proposed scheme capable of 104 Mbps encoding rate and test results are presented.