21st Century Computer Architecture

2y ago
134 Views
2 Downloads
3.23 MB
27 Pages
Last View : 17d ago
Last Download : 2m ago
Upload by : Julius Prosser
Transcription

21st Century Computer ArchitectureMark D. Hill, Univ. of Wisconsin-Madison6/2016 @ Architecture 2030 @ ISCA1. Whitepaper Content (tell a story)2. Process, Impact & Computing Community Consortium(CCC)

21st CenturyComputer ArchitectureA CCC community white tecturewhitepaper.pdf Information & Commun. Tech’s ImpactSemiconductor Technology’s ChallengesComputer Architecture’s FuturePre-Competitive Research JustifiedWas 21st Century Computer Architecture NEW(even in 2012)?

Our Task: Telling a “New” Story Was 21st Century Computer Architecture NEW?o Noo Resulted in more 50M in funding in USA3

In Communication of the China ComputerFederation (CCF)4

Our Task: Telling a “New” Story Was 21st Century Computer Architecture NEW?o Noo Resulted in more 50M in funding in USA Why?o New to otherso Told a story that mattered to others Develop a New-to-Others Message as a Storyo Why Important to Others?o Why Now?o How Might Research Make a Difference? This is our task now – not developing ideas new to us5

20th Century ICT Set Up Information & Communication Technology (ICT)Has Changed Our Worldo long list omitted Required innovations in algorithms, applications,programming languages, , & system software Key (invisible) enablers (cost-)performance gainso Semiconductor technology (“Moore’s Law”)o Computer architecture ( 80x per Danowitz et al.)6

Enablers: Technology ArchitectureArchitectureTechnologyDanowitz et al., CACM 04/2012, Figure 17

21st Century ICT Promises MoreData-centric personalized health careHuman network analysisComputation-driven scientific discoveryMuch more: known & unknown8

21st Century App CharacteristicsBIG DATASECURE/PRIVATEALWAYS ONLINEWhither enablers of future(cost-)performance gains? 9

Technology’s Challenges 1/2Late 20th CenturyThe New RealityMoore’s Law —2 transistors/chipTransistor count still 2 BUT Dennard Scaling — constant power/chipGone. Can’t repeatedly doublepower/chip10

Technology’s Challenges 2/2Late 20th CenturyThe New RealityMoore’s Law —2 transistors/chipTransistor count still 2 BUT Dennard Scaling — constant power/chipGone. Can’t repeatedly doublepower/chipModest (hidden)transistor unreliabilityIncreasing transistor unreliabilitycan’t be hiddenFocus on computationover communicationCommunication (energy) moreexpensive than computation1-time costs amortizedvia mass marketOne-time cost much worse &want specialized platformsHow should architects step up as technology falters?11

21st Century Comp Architecture20th Century21st CenturySingle-chip ingenericcomputerArchitecture as Infrastructure:Spanning sensors to cloudsXPerformance security, privacy,availability, programmability, Performancevia invisibleinstr.-levelparallelismEnergy First ParallelismX Specialization Cross-layer designPredictabletechnologies:CMOS, DRAM,& disksNew technologies (non-volatilememory, near-threshold, 3D,photonics, ) Rethink: memory &storage, reliability, communicationCrossCutting:Breakcurrentlayers withnewinterfaces12

21st Century Comp Architecture20th Century21st CenturySingle-chip ingenericcomputerArchitecture as Infrastructure:Spanning sensors to cloudsPerformance security, privacy,availability, programmability, Performancevia invisibleinstr.-levelparallelismEnergy First ParallelismX Specialization Cross-layer designPredictabletechnologies:CMOS, DRAM,& disksNew technologies (non-volatilememory, near-threshold, 3D,photonics, ) Rethink: memory &storage, reliability, communicationCrossCutting:Breakcurrentlayers withnewinterfaces13

21st Century Comp Architecture20th Century21st CenturySingle-chip ingenericcomputerArchitecture as Infrastructure:Spanning sensors to cloudsXPerformance security, privacy,availability, programmability, Performancevia invisibleinstr.-levelparallelismEnergy First ParallelismX Specialization Cross-layer designPredictabletechnologies:CMOS, DRAM,& disksNew technologies (non-volatilememory, near-threshold, 3D,photonics, ) Rethink: memory &storage, reliability, communicationCrossCutting:Breakcurrentlayers withnewinterfaces14

21st Century Comp Architecture20th Century21st CenturySingle-chip ingenericcomputerArchitecture as Infrastructure:Spanning sensors to cloudsXPerformance security, privacy,availability, programmability, Performancevia invisibleinstr.-levelparallelismEnergy First ParallelismX Specialization Cross-layer designPredictabletechnologies:CMOS, DRAM,& disksNew technologies (non-volatilememory, near-threshold, 3D,photonics, ) Rethink: memory &storage, reliability, communicationCrossCutting:Breakcurrentlayers withnewinterfaces15

21st Century Comp Architecture20th Century21st CenturySingle-chip instand-alonecomputerArchitecture as Infrastructure:Spanning sensors to cloudsPerformance security, privacy,availability, programmability, Performancevia invisibleinstr.-levelparallelismEnergy First Parallelism Specialization Cross-layer designPredictabletechnologies:CMOS, DRAM,& disksNew technologies (non-volatilememory, near-threshold, 3D,photonics, ) Rethink: memory &storage, reliability, communicationCrossCutting:Breakcurrentlayers withnewinterfaces16

What Research Exactly? Research areas in white paper (& backup slides)1.2.3.4.Architecture as Infrastructure: Spanning Sensors to CloudsEnergy FirstTechnology Impacts on ArchitectureCross-Cutting Issues & Interfaces Much more research developed by future PIs!17

Pre-Competitive Research Justified Retain (cost-)performance enabler to ICT revolution Successful companies cannot do this by themselveso Lack needed long-term focuso Don’t want to pay for what benefits allo Resist transcending interfaces that define their products Corroborateso Future of Computing Performance: Game Over or NextLevel?, National Academy Press, 2011o DARPA/ISAT Workshop Advancing Computer Systemswithout Technology Progress with outbrief http://www.cs.wisc.edu/ markhill/papers/isat2012 ACSWTP.pdf18

System Capability (log)“Timeline” from DARPA ISATFocruOusweNgyolohncTeSOCMFallow Period80s90s00s10s20s30s40s50sSource: Advancing Computer Systems without Technology Progress,ISAT Outbrief (http://www.cs.wisc.edu/ markhill/papers/isat2012 ACSWTP.pdf)Mark D. Hill and Christos Kozyrakis, DARPA/ISAT Workshop, March 26-27, 2012.Approved for Public Release, Distribution UnlimitedThe views expressed are those of the author and do not reflect the official policy or position of the Department19of Defense or the U.S. Government.

21st Century Computer ArchitectureMark D. Hill, Univ. of Wisconsin-Madison6/2016 @ Architecture 2030 @ ISCA1. Whitepaper Content (tell a story)2. Process, Impact & Computing CommunityConsortium (CCC)

White Paper ParticipantsSarita Adve, U Illinois *José F. Martínez, Cornell UDavid H. Albonesi, Cornell UMargaret Martonosi, Princeton U *David Brooks, Harvard UKunle Olukotun, Stanford ULuis Ceze, U Washington *Mark Oskin, U WashingtonSandhya Dwarkadas, U RochesterLi-Shiuan Peh, M.I.T.Joel Emer, Intel/MITMilos Prvulovic, Georgia TechBabak Falsafi, EPFLSteven K. Reinhardt, AMDAntonio Gonzalez, Intel/UPCMichael Schulte, AMD/U WisconsinMark D. Hill, U Wisconsin *,**Simha Sethumadhavan, Columbia UMary Jane Irwin, Penn State U *Guri Sohi, U WisconsinDavid Kaeli, Northeastern U *Daniel Sorin, Duke UStephen W. Keckler, NVIDIA/U TexasJosep Torrellas, U Illinois *Christos Kozyrakis, Stanford UThomas F. Wenisch, U Michigan *Alvin Lebeck, Duke UDavid Wood, U Wisconsin *MiloU PennsylvaniaKatherineYelick, UC Berkeley/LBNL *“*” Martin,contributedprose; “**” effortcoordinatorThanks of CCC, Erwin Gianchandani & Ed Lazowska forguidance and Jim Larus & Jeannette Wing for feedback 21

White Paper Process Late March 2012o CCC contacts coordinator & forms group April 2012ooooBrainstorm (meetings/online doc)Read related docs (PCAST, NRC Game Over, ACAR1/2, )Use online doc for intro & outline then parallel sectionsRotated authors to revise sections May 2012ooooBrainstorm list of researcher in/out of comp. architectureSolicit researcher feedback/endorsementDo distributed revision & redo of introRelease May 25 to CCC & via email Latero CCC & NSF Outbriefso HPCA/PPoPP/CGO/ICS Keynoteso ASPLOS & ISCA-Workshop Panels22

15M NSF XPS 2/2013 15Mfor 2/2014 later years23

Catalyzing and Enabling: Architecture2010Josep TorrellasUIUC2010Mark OskinWashington2012Mark HillWisconsinhttp://cra.org/ccc2013

Computing Community Consortium25CS eficiaries

CCC & You CCC Chair Greg Hager à Beth Mynatt Vice Chair Beth Mynatt à Mark Hill 20 Council Members Standing committee of Computing Research Association CRA established via NSF cooperative agreement You Be a rain-maker & give forward White papers Visioning And morehttp://cra.org/ccc26

Our Task: Telling a “New” Story Was 21st Century Computer Architecture NEW?o Noo Resulted in more 50M in funding Why?o New to otherso Told a story that mattered to others Develop a New-to-Others Message as a Storyo Why Important to Others?o Why Now?o How Might Research Make a Difference? This is our task now – not developing ideas new to us27

21st Century Comp Architecture 20th Century 21st Century Single-chip in generic computer Architecture as Infrastructure: Spanning sensors to clouds Performance security, privacy, availability, programmability, Cross-Cutting: Break current layers with new interfaces Performance via in

Related Documents:

What is Computer Architecture? “Computer Architecture is the science and art of selecting and interconnecting hardware components to create computers that meet functional, performance and cost goals.” - WWW Computer Architecture Page An analogy to architecture of File Size: 1MBPage Count: 12Explore further(PDF) Lecture Notes on Computer Architecturewww.researchgate.netComputer Architecture - an overview ScienceDirect Topicswww.sciencedirect.comWhat is Computer Architecture? - Definition from Techopediawww.techopedia.com1. An Introduction to Computer Architecture - Designing .www.oreilly.comWhat is Computer Architecture? - University of Washingtoncourses.cs.washington.eduRecommended to you b

concept “21st OR twenty-first century skills”, this has, also, been searched through the terms: “21st OR twenty-first century competencies”, “21st OR twenty-first century literacy”, “21st OR twenty-first century learn*”. The search action returned 116 results. The references sections of these texts were

To understand the 21st Century Life and Career Standards To gain knowledge in the Framework for 21st Century Inter-disciplinary themes To develop skills in planning cross-curricular activities related to the 21st Century skills To learn practical strategies in how to infuse 21st Century skills

Paper Name: Computer Organization and Architecture SYLLABUS 1. Introduction to Computers Basic of Computer, Von Neumann Architecture, Generation of Computer, . “Computer System Architecture”, John. P. Hayes. 2. “Computer Architecture and parallel Processing “, Hwang K. Briggs. 3. “Computer System Architecture”, M.Morris Mano.

with the definition, demand and delivery of 21st Century Skills, with a particular focus on low- and middle-income countries (LMICs). In looking at definitions of ‘21st Century Skills’, Section 2 of this study finds that there is a broad range of available literature discussing 21st Century Skills, including a number of key synthesis studies.

“21st century skills” and “college and career readiness” have recently become watchwords in education. The Partnership for 21st century skills advocates adoption of local, state, and federal policies that support explicit integration of 21st century skills into instruction for all students (P21, 2009).

districts or state departments of education implementing 21st century skills now rely on the Partnership for 21st Century Skills’ list or some variation of it, though there are exceptions. We also identified one up-and-coming list, the Assessment and Teaching of 21st Century Skills (ATC21S) initiative, sponsored by Microsoft and Cisco Systems.

The Baldrige Excellence Framework empowers your organization to reach its goals, improve results, and become more competitive. Thousands of organizations around the world use the Baldrige Excellence Framework to improve and get sustainable results. Those recognized as national role models receive the Malcolm Baldrige National Quality Award, a Presidential award. At the Quality Texas Foundation .