The Von Neumann Machine X86-64 Overview 1 - Courses.cs.vt.edu

The von Neumann MachineX86-64 Overview 11945: John von Neumann Wrote a report on the stored program concept, known as the First Draft of a Report onEDVACalso Alan Turing Konrad Zuse Eckert & Mauchly The basic structure proposed in the draft became known as the “von Neumann machine”(or model). CS@VTa memory, containing instructions and dataa processing unit, for performing arithmetic and logical operationsa control unit, for interpreting instructionsComputer Organization I 2005-2020 WD McQuain

The von Neumann MachineX86-64 Overview 2stores both programinstructions and dataMEMORYMARMDRDataPROCESSING ssDataaddressControlsignalsdecodes current instruction,CONTROL UNITPCprogram counter: points tothe next instruction to befetchedCS@VTmanages processing unit tocarry out instructionIRinstruction register: storescurrent instructionComputer Organization I 2005-2020 WD McQuain

Intel x86 FamilyX86-64 Overview 3Totally dominate laptop/desktop/server marketEvolutionary design Backwards compatible to the 8086, introduced in 1978Open architecture: 3rd party suppliers for all sorts of external hardware andsoftwareAdded more features as time went onComplex instruction set computer architecture (CISC) Many different instructions with many different formats Hard to match performance of Reduced Instruction Set Architectures (RISC)But, Intel has done just that! CS@VTBut, only small subset encountered with Linux programsIn terms of speed. Less so for low power.Computer Organization I 2005-2020 WD McQuain

Intel x86 HistoryName8086 CS@VTX86-64 Overview 4Date1978Transistors29KClock rate5-10 MHzFirst 16-bit processor. Basis for IBM PC & DOS1MB address space8088: slight modification of 8086 chip used in IBM PCQuickly dominated personal computer market in sales via IBM PCComputer Organization I 2005-2020 WD McQuain

Intel x86 HistoryName80386 CS@VTX86-64 Overview 5Date1985Transistors275KClock rate16-33 MHzFirst 32 bit processor , referred to as IA32 architectureAdded “flat addressing”Capable of running Unix32-bit Linux/gcc used no instructions introduced in later modelsIBM PC superceded by 386-based machines from other vendors; explosion of "clone"vendorsComputer Organization I 2005-2020 WD McQuain

Intel x86 HistoryX86-64 Overview 6Intel Attempted Radical Shift from IA32 to IA64 Totally different architecture (Itanium)Executes IA32 code only as legacyPerformance disappointingAMD Stepped in with Evolutionary Solution x86-64 (now called “AMD64”)Intel Felt Obligated to Focus on IA64 Hard to admit mistake or that AMD's approach is better2004: Intel Announces EM64T extension to IA32 Extended Memory 64-bit TechnologyAlmost identical to x86-64! (Earlier Intel-AMD licensing agreement gives rights.)All but low-end x86 processors support x86-64 CS@VTBut, lots of code still runs in 32-bit modeComputer Organization I 2005-2020 WD McQuain

Intel x86 HistoryNamePentium 4F CS@VTX86-64 Overview 7Date2004Transistors125MClock rate2800-3800 MHzFirst (successful) 64-bit processor (from Intel), referred to as x86-64Based on AMD 64-bit architecture, which was derived from IA-32Computer Organization I 2005-2020 WD McQuain

The Power WallX86-64 Overview 8"Shortly after the beginning of the 21st century an inflection point was reached: in aspeech in 2001, and Intel executive pointed out that an extrapolation of increasing powerdensity (and thus temperature) of microprocessor chips would exceed that of a rocketnozzle by 2006, and of the sun's surface by 2012.""Clearly something had to give, and the result is that going forward we can expect anexponential increase in the number of processors (cores) available to us, but not in theirindividual performance."CS@VTComputer Organization I 2005-2020 WD McQuain

Core I7: the multicore revolutionX86-64 Overview 9Response to the Power Wall CS@VTpackage multiple cores on a single processor diecompensate for stagnant clock rates with parallel computation2, 4, 6, 8 cores and beyondComputer Organization I 2005-2020 WD McQuain

Intel x86 HistoryNameCore i7 CS@VTX86-64 Overview 10Date2008Transistors731MClock rate2667-3333 MHzMulti-core architecture (4, 6, 8 cores)Reduced clock rates compensated for by multiple coresComputer Organization I 2005-2020 WD McQuain

The von Neumann MachineX86-64 Overview 11MEMORYMEMORYMARMDRDataProgram dataPROCESSING ssDataaddressProgram object codeControlsignalsOS object codeOS dataRuntime stackCONTROL UNITPCCS@VTIRComputer Organization I 2005-2020 WD McQuain

The von Neumann MachineProgrammer-visible State:X86-64 Overview 12PROCESSING UNITREGISTERSPC: program counter stores address of next instructioncalled EIP (IA32) or RIP (x86-64)ALUCONDITION CODESRegister file: stores heavily used program dataCONTROL UNITCondition codes CS@VTstore status information about most recentarithmetic-logical operationused for conditional branchingComputer Organization IPCIR 2005-2020 WD McQuain

IA32 Integer RegistersX86-64 Overview 13PROCESSING UNIT16-bit virtual registers(backwards compatibility)REGISTERSALUgeneral purposeCONDITION ebx%bx%bh%bl%esi%si%edi%di%esp%sp%ebp%bpComputer Organization I 2005-2020 WD McQuain

x86-64 Integer RegistersX86-64 Overview d existing registers. Add 8 new ones.Make %ebp/%rbp general purposeComputer Organization I 2005-2020 WD McQuain

Programming the MachineX86-64 Overview 15// C source codeint imax(int first, int second) {if ( first second )return first;return second;}But the hardware only"understands" binaryrepresentationsCS@VTComputer Organization I 2005-2020 WD McQuain

Programming the MachineX86-64 Overview 16int imax(int first, int second) {if ( first second )return first;return second;}gcc-O0 -c –m64 -std c99 20352e322e382d62756e7575742934. . .But who wants toprogram in binary?CS@VTComputer Organization I 2005-2020 WD McQuain

Programming the MachineX86-64 Overview 17int imax(int first, int second) {if ( first second )return first;return second;. . .imax:}Solution:translate high-level language codeinto intermediate-level code which ismore human-friendly,then translate that "assembly" codeinto the machine's p%rsp, %rbp 8, %rsp%edi, -4(%rbp)%esi, -8(%rbp)-4(%rbp), %eax-8(%rbp), %eax.L2-4(%rbp), %eax.L3movl-8(%rbp), %eaxpopqret%rbp.L2:.L3:. . .CS@VTComputer Organization I 2005-2020 WD McQuain

CS@VT 2005-2020 WD McQuain The Power Wall "Shortly after the beginning of the 21st century an inflection point was reached: in a speech in 2001, and Intel executive pointed out that an extrapolation of increasing power density (and thus temperature) of microprocessor chips would exceed that of a rocket