Chapter 1 Fundamentals of Quantitative Design and Analysis Computer Architecture A Quantitative Approach Sixth Edition Computer Technology Performance improvements Improvements in semiconductor technology ID: 776392
Download Presentation The PPT/PDF document " Copyright © 2019, Elsevier Inc. All ri..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Copyright © 2019, Elsevier Inc. All rights reserved.
Chapter 1
Fundamentals of Quantitative Design and Analysis
Computer Architecture
A Quantitative Approach
, Sixth Edition
Slide2Computer Technology
Performance improvements:Improvements in semiconductor technologyFeature size, clock speedImprovements in computer architecturesEnabled by HLL compilers, UNIXLead to RISC architecturesTogether have enabled:Lightweight computersProductivity-based managed/interpreted programming languages
Copyright © 2019, Elsevier Inc. All rights reserved.
Introduction
Slide3Single Processor Performance
Copyright © 2019, Elsevier Inc. All rights reserved.
Introduction
Slide4Copyright © 2019, Elsevier Inc. All rights reserved.
Current Trends in Architecture
Cannot continue to leverage Instruction-Level parallelism (ILP)Single processor performance improvement ended in 2003New models for performance:Data-level parallelism (DLP)Thread-level parallelism (TLP)Request-level parallelism (RLP)These require explicit restructuring of the application
Introduction
Slide5Copyright © 2019, Elsevier Inc. All rights reserved.
Classes of Computers
Personal Mobile Device (PMD)e.g. start phones, tablet computersEmphasis on energy efficiency and real-timeDesktop ComputingEmphasis on price-performanceServersEmphasis on availability, scalability, throughputClusters / Warehouse Scale ComputersUsed for “Software as a Service (SaaS)”Emphasis on availability and price-performanceSub-class: Supercomputers, emphasis: floating-point performance and fast internal networksInternet of Things/Embedded ComputersEmphasis: price
Classes of Computers
Slide6Copyright © 2019, Elsevier Inc. All rights reserved.
Parallelism
Classes of parallelism in applications:Data-Level Parallelism (DLP)Task-Level Parallelism (TLP)Classes of architectural parallelism:Instruction-Level Parallelism (ILP)Vector architectures/Graphic Processor Units (GPUs)Thread-Level ParallelismRequest-Level Parallelism
Classes of Computers
Slide7Copyright © 2019, Elsevier Inc. All rights reserved.
Flynn’s Taxonomy
Single instruction stream, single data stream (SISD)Single instruction stream, multiple data streams (SIMD)Vector architecturesMultimedia extensionsGraphics processor unitsMultiple instruction streams, single data stream (MISD)No commercial implementationMultiple instruction streams, multiple data streams (MIMD)Tightly-coupled MIMDLoosely-coupled MIMD
Classes of Computers
Slide8Copyright © 2019, Elsevier Inc. All rights reserved.
Defining Computer Architecture
“Old” view of computer architecture:Instruction Set Architecture (ISA) designi.e. decisions regarding:registers, memory addressing, addressing modes, instruction operands, available operations, control flow instructions, instruction encoding“Real” computer architecture:Specific requirements of the target machineDesign to maximize performance within constraints: cost, power, and availabilityIncludes ISA, microarchitecture, hardware
Defining Computer Architecture
Slide9Instruction Set Architecture
Class of ISAGeneral-purpose registersRegister-memory vs load-storeRISC-V registers32 g.p., 32 f.p.
Copyright © 2019, Elsevier Inc. All rights reserved.
Defining Computer Architecture
RegisterNameUseSaverx0zeroconstant 0n/ax1rareturn addrcallerx2spstack ptrcalleex3gpgbl ptrx4tpthread ptrx5-x7t0-t2temporariescallerx8s0/fpsaved/frame ptrcallee
Register
Name
Use
Saver
x9
s1
saved
callee
x10-x17
a0-a7
arguments
caller
x18-x27
s2-s11
saved
callee
x28-x31
t3-t6
temporaries
caller
f0-f7
ft0-ft7
FP temps
caller
f8-f9
fs0-fs1
FP saved
callee
f10-f17
fa0-fa7
FP arguments
callee
f18-f27
fs2-fs21
FP saved
callee
f28-f31
ft8-ft11
FP temps
caller
Slide10Instruction Set Architecture
Memory addressingRISC-V: byte addressed, aligned accesses fasterAddressing modesRISC-V: Register, immediate, displacement (base+offset)Other examples: autoincrement, indexed, PC-relativeTypes and size of operandsRISC-V: 8-bit, 32-bit, 64-bit
Copyright © 2019, Elsevier Inc. All rights reserved.
Defining Computer Architecture
Slide11Instruction Set Architecture
OperationsRISC-V: data transfer, arithmetic, logical, control, floating pointSee Fig. 1.5 in textControl flow instructionsUse content of registers (RISC-V) vs. status bits (x86, ARMv7, ARMv8)Return address in register (RISC-V, ARMv7, ARMv8) vs. on stack (x86)EncodingFixed (RISC-V, ARMv7/v8 except compact instruction set) vs. variable length (x86)
Copyright © 2019, Elsevier Inc. All rights reserved.
Defining Computer Architecture
Slide12Copyright © 2019, Elsevier Inc. All rights reserved.
Trends in Technology
Integrated circuit technology (Moore’s Law)Transistor density: 35%/yearDie size: 10-20%/yearIntegration overall: 40-55%/yearDRAM capacity: 25-40%/year (slowing)8 Gb (2014), 16 Gb (2019), possibly no 32 GbFlash capacity: 50-60%/year8-10X cheaper/bit than DRAMMagnetic disk capacity: recently slowed to 5%/yearDensity increases may no longer be possible, maybe increase from 7 to 9 platters8-10X cheaper/bit then Flash200-300X cheaper/bit than DRAM
Trends in Technology
Slide13Copyright © 2019, Elsevier Inc. All rights reserved.
Bandwidth and Latency
Bandwidth or throughputTotal work done in a given time32,000-40,000X improvement for processors300-1200X improvement for memory and disksLatency or response timeTime between start and completion of an event50-90X improvement for processors6-8X improvement for memory and disks
Trends in Technology
Slide14Copyright © 2019, Elsevier Inc. All rights reserved.
Bandwidth and Latency
Log-log plot of bandwidth and latency milestones
Trends in Technology
Slide15Copyright © 2019, Elsevier Inc. All rights reserved.
Transistors and Wires
Feature sizeMinimum size of transistor or wire in x or y dimension10 microns in 1971 to .011 microns in 2017Transistor performance scales linearlyWire delay does not improve with feature size!Integration density scales quadratically
Trends in Technology
Slide16Copyright © 2019, Elsevier Inc. All rights reserved.
Power and Energy
Problem: Get power in, get power outThermal Design Power (TDP)Characterizes sustained power consumptionUsed as target for power supply and cooling systemLower than peak power (1.5X higher), higher than average power consumptionClock rate can be reduced dynamically to limit power consumptionEnergy per task is often a better measurement
Trends in Power and Energy
Slide17Copyright © 2019, Elsevier Inc. All rights reserved.
Dynamic Energy and Power
Dynamic energyTransistor switch from 0 -> 1 or 1 -> 0½ x Capacitive load x Voltage2Dynamic power½ x Capacitive load x Voltage2 x Frequency switchedReducing clock rate reduces power, not energy
Trends in Power and Energy
Slide18Copyright © 2019, Elsevier Inc. All rights reserved.
Power
Intel 80386 consumed ~ 2 W3.3 GHz Intel Core i7 consumes 130 WHeat must be dissipated from 1.5 x 1.5 cm chipThis is the limit of what can be cooled by air
Trends in Power and Energy
Slide19Copyright © 2019, Elsevier Inc. All rights reserved.
Reducing Power
Techniques for reducing power:Do nothing wellDynamic Voltage-Frequency ScalingLow power state for DRAM, disksOverclocking, turning off cores
Trends in Power and Energy
Slide20Copyright © 2019, Elsevier Inc. All rights reserved.
Static Power
Static power consumption25-50% of total powerCurrentstatic x VoltageScales with number of transistorsTo reduce: power gating
Trends in Power and Energy
Slide21Copyright © 2019, Elsevier Inc. All rights reserved.
Trends in Cost
Cost driven down by learning curveYieldDRAM: price closely tracks costMicroprocessors: price depends on volume10% less for each doubling of volume
Trends in Cost
Slide22Copyright © 2019, Elsevier Inc. All rights reserved.
Integrated Circuit Cost
Integrated circuitBose-Einstein formula:Defects per unit area = 0.016-0.057 defects per square cm (2010)N = process-complexity factor = 11.5-15.5 (40 nm, 2010)
Trends in Cost
Slide23Copyright © 2019, Elsevier Inc. All rights reserved.
Dependability
Module reliabilityMean time to failure (MTTF)Mean time to repair (MTTR)Mean time between failures (MTBF) = MTTF + MTTRAvailability = MTTF / MTBF
Dependability
Slide24Copyright © 2019, Elsevier Inc. All rights reserved.
Measuring Performance
Typical performance metrics:Response timeThroughputSpeedup of X relative to YExecution timeY / Execution timeXExecution timeWall clock time: includes all system overheadsCPU time: only computation timeBenchmarksKernels (e.g. matrix multiply)Toy programs (e.g. sorting)Synthetic benchmarks (e.g. Dhrystone)Benchmark suites (e.g. SPEC06fp, TPC-C)
Measuring Performance
Slide25Copyright © 2019, Elsevier Inc. All rights reserved.
Principles of Computer Design
Take Advantage of Parallelisme.g. multiple processors, disks, memory banks, pipelining, multiple functional unitsPrinciple of LocalityReuse of data and instructionsFocus on the Common CaseAmdahl’s Law
Principles
Slide26Copyright © 2019, Elsevier Inc. All rights reserved.
Principles of Computer Design
The Processor Performance Equation
Principles
Slide27Copyright © 2019, Elsevier Inc. All rights reserved.
Principles of Computer Design
Principles
Different instruction types having different CPIs
Slide28Copyright © 2019, Elsevier Inc. All rights reserved.
Principles of Computer Design
Principles
Different instruction types having different CPIs
Slide29Fallacies and Pitfalls
All exponential laws must come to an endDennard scaling (constant power density)Stopped by threshold voltageDisk capacity30-100% per year to 5% per yearMoore’s LawMost visible with DRAM capacityITRS disbandedOnly four foundries left producing state-of-the-art logic chips11 nm, 3 nm might be the limit
Copyright © 2019, Elsevier Inc. All rights reserved.
Slide30Fallacies and Pitfalls
Microprocessors are a silver bulletPerformance is now a programmer’s burdenFalling prey to Amdahl’s LawA single point of failureHardware enhancements that increase performance also improve energy efficiency, or are at worst energy neutralBenchmarks remain valid indefinitelyCompiler optimizations target benchmarks
Copyright © 2019, Elsevier Inc. All rights reserved.
Slide31Fallacies and Pitfalls
The rated mean time to failure of disks is 1,200,000 hours or almost 140 years, so disks practically never failMTTF value from manufacturers assume regular replacementPeak performance tracks observed performanceFault detection can lower availabilityNot all operations are needed for correct execution
Copyright © 2019, Elsevier Inc. All rights reserved.