ECE/CS 752: Advanced Computer Architecture IComputer ArchitectureTechnologySemiconductor HistoryComputer Architecture752 In ContextWhy Take 752?Slide 8Technology PushPerformance GrowthSlide 11Application PullSlide 13TrendsDynamic PowerMulticore ManiaWhy MulticoreAmdahl’s LawFixed Chip Power BudgetFixed Power ScalingFocus of this CourseInstruction Set ProcessingIron LawSlide 24Our GoalTextbooksExpected BackgroundCourse ContextAbout This CourseSlide 30Slide 31Slide 32Slide 33Slide 34Slide 35Tentative ScheduleWrapping UpECE/CS 752: Advanced Computer Architecture IInstructor:Mikko H LipastSpring 2012University of Wisconsin-MadisonLecture notes based on slides created by John Shen, Mark Hill, David Wood, Guri Sohi, and Jim Smith•Rely on abstraction layers to manage complexity–Von Neumann MachineApplicationsTechnologyComputerArchitectureComputer ArchitectureQuantum PhysicsTransistors & DevicesLogic Gates & MemoryVon Neumann Machinex86 Machine PrimitvesVisual C++Firefox, MS ExcelWindows 7Technology•Technology advances at astounding rate–19th century: attempts to build mechanical computers–Early 20th century: mechanical countng systems (cash registers, etc.)–Mid 20th century: vacuum tubes as switches–Since: transistors, integrated circuits•1965: Moore’s law [Gordon Moore]–Predicted doubling of IC capacity every 18 months–Has held and will contnue to hold•Drives functonality, performance, cost–Exponental improvement for 40+ yearsSemiconductor HistoryDate Event Comments1947 1st transistor Bell Labs1958 1st IC Jack Kilby (MSEE ’50) @TIWinner of 2000 Nobel prize1971 1st microprocessor Intel (calculator market)1974 Intel 4004 2300 transistors1978 Intel 8086 29K transistors1989 Intel 80486 1M transistors1995 Intel Pentium Pro 5.5M transistors2006 Intel Montecito 1.7B transistors201x IBM 50B transistorsComputer Architecture •Instructon Set Architecture (IBM 360)–… the attributes of a [computing] system as seen by the programmer. I.e. the conceptual structure and functional behavior, as distinct from the organization of the data flows and controls, the logic design, and the physical implementation. -- Amdahl, Blaaw, & Brooks, 1964•Machine Organizaton (microarchitecture)–ALUS, Buses, Caches, Memories, etc.•Machine Implementaton (realizaton)–Gates, cells, transistors, wires752 In Context•Prior courses–352 – gates up to multplexors and adders–354 – high-level language down to machine language interface or instructon set architecture (ISA)–552 – implement logic that provides ISA interface–CS 537 – provides OS background (co-req. OK)•This course – 752 – covers advanced techniques–Modern processors that exploit ILP–Modern memory systems that exploit MLP•Additonal courses–ECE 757 covers parallel and multprocessing–ECE 755 covers VLSI designWhy Take 752?•To become a computer designer–Alumni of this class helped design your computer•To learn what is under the hood of a computer–Innate curiosity–To better understand when things break–To write better code/applicatons–To write better system software (O/S, compiler, etc.)•Because it is intellectually fascinatng!–What is the most complex man-made single device?Computer Architecture•Exercise in engineering tradeoff analysis–Find the fastest/cheapest/power-efficient/etc. soluton–Optmizaton problem with 100s of variables•All the variables are changing–At non-uniform rates–With inflecton points–Only one guarantee: Today’s right answer will be wrong tomorrow•Two high-level effects:–Technology push–Applicaton PullTechnology Push•What do these two intervals have in common?–1776-1999 (224 years)–2000-2001 (2 years)•Answer: Equal progress in processor speed!•The power of exponental growth!•Driven by Moore’s Law•Devices per chip doubles every 18-24 months•Computer architects turn additonal resources into•Speed•Power savings•FunctonalityPerformance GrowthUnmatched by any other industry ![John Crawford, Intel]•Doubling every 18 months (1982-1996): 800x–Cars travel at 44,000 mph and get 16,000 mpg–Air travel: LA to NY in 22 seconds (MACH 800)–Wheat yield: 80,000 bushels per acreDoubling every 24 months (1971-1996): 9,000x–Cars travel at 600,000 mph, get 150,000 mpg–Air travel: LA to NY in 2 seconds (MACH 9,000)–Wheat yield: 900,000 bushels per acreTechnology Push•Technology advances at varying rates–E.g. DRAM capacity increases at 60%/year–But DRAM speed only improves 10%/year–Creates gap with processor frequency!•Inflecton points–Crossover causes rapid change–E.g. enough devices for multcore processor (2001)•Current issues causing an “inflecton point”–Power consumpton–Reliability–VariabilityApplicaton Pull•Corollary to Moore’s Law: Cost halves every two yearsIn a decade you can buy a computer for less than its sales tax today. –Jim Gray•Computers cost-effectve for–Natonal security – weapons design–Enterprise computng – banking–Departmental computng – computer-aided design–Personal computer – spreadsheets, email, web–Mobile computng – GPS, locaton-aware, ubiquitousApplicaton Pull•What about the future?–E.g. weather forecastng computatonal demand•Must dream up applicatons that are not cost-effectve today–Virtual reality, telepresence–Web agents, social networking–Wireless, locaton-aware–Proactve (beyond interactve) w/ sensors–Recogniton/Mining/Synthesis (RMS)–???•This is your job!Trends•Moore’s Law for device integraton•Chip power consumpton•Single-thread performance trend[source: Intel]Dynamic Power•Statc CMOS: current flows when actve–Combinatonal logic evaluates new inputs–Flip-flop, latch captures new value (clock edge)•Terms–C: capacitance of circuit•wire length, number and size of transistors–V: supply voltage–A: actvity factor–f: frequency•Future: Fundamentally power-constrainedunitsiiiidynfAVCkP2Sep 18, 2007 Mikko Lipasti-University of WisconsinSep 18, 2007Mikko Lipasti-University of WisconsinMultcore Mania•First, servers–IBM Power4, 2001•Then desktops–AMD Athlon X2, 2005•Then laptops–Intel Core Duo, 2006•Your cellphone–Baseband/DSP/applicaton/graphicsWhy MultcoreSingle Core Dual Core Quad CoreCore area A ~A/2 ~A/4Core power W ~W/2 ~W/4Chip power W + O W + O’ W + O’’Core performance P 0.9P 0.8PChip performance P 1.8P 3.2PSep 18, 2007Mikko Lipasti-University of