Embedded System Memory Organization Overview Of Real-time And Embedded .
Embedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkData compression reviewRobert Dickhttp://robertdick.org/esds/Office: EECS 2417-EDepartment of Electrical Engineering and Computer ScienceUniversity of MichiganLossy compression.Lossless compression.Uses in embedded systems.Predictive models.Relationship with intelligence: Hutter Prize. 50,000 per percent.Kolmogorov complexity.3Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkRobert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsEmbedded system memory organizationCollaborators on projectControl of datum location: static or dynamic?Implications of real-time deadlines?Implications of networked systems?NEC Labs AmericaGanesh LakshminarayanaAnand RaghunathanPrincetonNiraj K. JhaImplications of tight constraints on transistor count?Implications of tight constraints on memory?Errors related to misuse of memory?5Robert DickEmbedded System Design and Synthesis8Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsIntroductionReal-time operating systems (RTOS)Interaction between HW and SWReal-Time Operating Systems are often used in embeddedsystemsRapid response to interruptsHW interface abstractionThey simplify use of hardware, ease management of multipletasks, and adhere to real-time constraintsInteraction between different tasksCommunicationSynchronizationPower is important in many embedded systems with RTOSsRTOSs can consume significant amount of powerMultitaskingThey are re-used in many embedded systemsIdeally fully preemptivePriority-based schedulingFast context switchingSupport for real-time clockThey impact power consumed by application softwareRTOS power effects influence system-level design10Robert DickEmbedded System Design and Synthesis11Robert DickEmbedded System Design and Synthesis
Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsGeneral-purpose OS stressData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsRTOSs stressPredictable service execution timesGood average-case behaviorPredictable schedulingProviding many servicesGood worst-case behaviorSupport for a large number of hardware devicesLow memory usageSpeedSimplicity12Robert DickEmbedded System Design and Synthesis13Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsPredictabilityRTOS overviewApplicationsMPEGencodingGeneral-purpose computer architecture focuses on lative -time embedded systems need predictabilityBasicIOISRTaskmanagerDisabling or locking caches is commonCareful evaluation of worst-case is essentialSpecialized or static memory management commonMicro browserProcessorTimerMemoryOther hardwareNetwork asks14Robert DickEmbedded System Design and Synthesis15Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsRTOS power consumptionRTOS and real-time referencesK. Ramamritham and J. Stankovic. Scheduling algorithms andoperating systems support for real-time systems. Proc. IEEE,82(1):55–67, January 1994Used in several low-power embedded systemsNeed for RTOS power analysisSignificant power consumptionImpacts application software powerRe-used across several applications16Robert DickEmbedded System Design and SynthesisGiorgio C. Buttazzo. Hard Real-Time Computing Systems.Kluwer Academic Publishers, Boston, 200017Robert DickEmbedded System Design and Synthesis
Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsPrior workData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsEmbedded OS power referencesVivek Tiwari, Sharad Malik, and Andrew Wolfe. Compilationtechniques for low energy: An overview. In Proc. Int. Symp.Low-Power Electronics, pages 38–39, October 1994Y. Li and J. Henkel. A framework for estimating and minimizingenergy dissipation of embedded HW/SW systems. In Proc.Design Automation Conf., pages 188–193, June 1998J. J. Labrosse. MicroC/OS-II. R & D Books, KS, 199818T. Cignetti, K. Komarov, and C. Ellis. Energy estimation toolsfor the Palm. In Proc. Int. Wkshp. on Modeling, Analysis andSimulation of Wireless and Mobile Systems, pages 96–103,August 2000.Robert P. Dick, G. Lakshminarayana, A. Raghunathan, andNiraj K. Jha. Analysis of power dissipation in real-time operatingsystems. IEEE Trans. Computer-Aided Design of IntegratedCircuits and Systems, 22(5):615–627, May 2003.A Shye, B Scholbrock, and G Memik. Into the wild: studing realuser activity patterns to guide power optimizations for mobilearchitectures. In Proc. Int. Symp. on Microarchitecture, pages168–178, 2009.M Dong and L Zhong. Sesame: A self-constructive virtual powermeter for battery-powered mobile systems. Technical report,2010.Robert DickEmbedded System Design and SynthesisL. Zhang, B. Tiwana, Z. Qian, Z. Wang, R. P. Dick, Z. M. Mao,and L. Yang. Accurate online power estimation and automaticbattery behavior based power model generation for smartphones.In Proc. Int. Conf. Hardware/Software Codesign and SystemSynthesis, pages 105–114, October 2010.Robert DickEmbedded System Design and Synthesis19Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkRTOS power referencesContributionsK. Baynes, C. Collins, E. Fiterman, B. Ganesh, P. Kohout,C. Smit, T. Zhang, and B. Jacob. The performance and energyconsumption of three embedded real-time operating systems. InProc. Int. Conf. Compilers, Architecture & Synthesis forEmbedded Systems, pages 203–210, November 2001First detailed power analysis of RTOSProof of concept later used by othersApplicationsLow-power RTOSEnergy-efficient software architectureIncorporate RTOS effects in system designT.-K. Tan, A. Raghunathan, and Niraj K. Jha. EMSIM: Anenergy simulation framework for an embedded operating system.In Proc. Int. Symp. Circuits & Systems, pages 464–467, May200220Introduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsRobert DickEmbedded System Design and Synthesis21Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsSimulated embedded systemPeriodically triggered ABSSense speed andpedal conditionsIBM0118160PT3 60DRAMFujitsuSPARClite 86832On chip cacheIBM0118160PT3 60DRAMTimerEPROMUARTLEDsYEasy to add new devicesTimertransition?Cycle-accurate modelInterruptsOther ASICsand peripheralsFujitsu board support libraryused in modelNµC/OS-II RTOS usedBrake decisionProcessorbus23SleepRobert DickEmbedded System Design and SynthesisComputeacceleration24Robert DickActuate brakeEmbedded System Design and Synthesis
Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsPeriodically triggered ABS timingData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsSelectively triggered ABSPedalpressed?TimerSense speed andpedal rocessBrake uate brakeYTime25Robert DickEmbedded System Design and Synthesis26Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsSelectively triggered ABS timingAgent exampleAgent 1TimerMoneyCommodity 1Commodity 2Commodity 3Commodity 4BrakepedalAdvertiseABSprocessAgent 3Agent 6BidAgent 2OfferWheelsensorTransfer resultsBrakeactionTimeAgent 563% reduction in energy and power consumption27Agent 4Robert DickEmbedded System Design and Synthesis28Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsSingle task network interfaceMulti-tasking network interfaceChecksumcomputationGet packetGet nsferpacketsReleaseEthernetcontrollerChecksum computationand outputRTOS power analysis suggests process re-organization.21% reduction in energy consumption. Similar power consumption.Procuring Ethernet controller has high energy cost29Robert DickEmbedded System Design and Synthesis30Robert DickEmbedded System Design and Synthesis
Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation Introduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsABS optimization effectsEnergy by calltree position fortask ASPARClitecompilerOS codeOSSched()main()ExternalstimulusSPARClite cachesimulatorTimermodelSPARClite ISSInstruction levelenergy modelUARTmodelCachecontrollermodelBusinterfaceunit modelEnergy by calltree position fortask BABSApplicationFloating k controlRedesigned application afterusing simulator to locateareas where power was wasted63% energy reduction63% power reductionRTOS directly accounted for50% of system energytegan notegaMemory modelMemoryenergy modelModels forotherperipheralsEnergy 0450040003500300025002000150010005000Robert DickEmbedded System Design and Synthesis34Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResults32Agent optimization effectsEthernet optimization effects375032503500ApplicationFloating k control3000Energy Mail version used RTOSmailboxes for informationtransmissionEthernet275025002250Energy (mJ)Agent3250Tuned version carefullyhand-tuned to used sharedmemoryPower can be reduced at acost200017501500125010007505002500fbun df35Determined thatsynchronization routine costwas highUsed RTOS buffering toamortize synchronizationcosts20.5% energy reduction0.2% power reductionRTOS directly accounted for1% of system energyEnergy savings due toimproved RTOS use, notreduced RTOS energybunonelaitumIncreased applicationsoftware complexityDecreased flexibilityApplicationFloating k controlRobert DickEmbedded System Design and Synthesis36Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsMailbox exampleSemaphore example375325Mailbox300Energy ing k controlEnergy (mJ)350Rapid mailbox communicationbetween tasksRTOS directly accounted for99% of system energy037Robert DickEmbedded System Design and ting k controlRobert DickSemaphores used for tasksynchronizationRTOS directly accounted for98.7% of system energyEmbedded System Design and Synthesis
Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkEnergy boundsIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsTime resultsService27501400013000250012000225011000Time ert DickMailbox SemaphoreTask1508.88 mJ total48.69 %init tvecsinit timer18.01 mJ total1.72 %startup7.39 mJ total0.71 %win unf trapOSDisableIntOSEnableIntOSSemPend104.59 mJ total10.01 %OSSemPost9.82 mJ total0.94 %OSTimeGet4.62 mJ total0.44 %CPUInit0.29 mJ total0.03 %printf368.07 mJ total35.22 %Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy eled1.314.26Energy (%)0.000.00Time (ms)0.000.00Calls11win unf .17115win unf o mainsave datainit datainit bsscache onIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsExample power-efficient change to RTOSSmall changes can greatly improve RTOS power consumptionµC/OS-II tracks processor loading by incrementing a counterwhen idleHowever, this is not a good low-power design decisionNOPs have lower power than add or increment instructionsSleep mode has much lower powerCan disable loading counter and use NOPs or sleep modeRobert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsRTOS ConclusionsAlternatively, can use timer-based samplingDemonstrated that RTOS significantly impacts powerNormally NOP or sleep when idleWake up on timer ticksSample highest non-timer ISR taskIf it’s the idle task, increment a counterCan dramatically reduce power consumption without losingfunctionality43Robert DickEmbedded System Design and SynthesisIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsExample power-efficient change to RTOS42Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkMaximumenergy .372.5320347.0065.651.311.31···Embedded System Design and SynthesisSemaphore example hierarchical call treerealstart25.40 mJ total2.43 %AgentTaskfptodpBSPInitfstatCPUInitfstat rGetPsrinit bssGetTbrinit dataInitTimerinit timerOSCtxSwinit tvecsOSDisableInt···ApplicationFloating k controlbun dteAgent39100nolneaitumgan 01005002752500Minimumenergy 2.5318012.1046.631.310.84···RTOS power analysis can improve application software designApplicationsLow-power RTOS designEnergy-efficient software architectureConsider RTOS effects during system design44Robert DickEmbedded System Design and Synthesis
Data compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkIntroduction, motivation, and past workExamples of energy optimizationSimulation infrastructureResultsReferenceData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkMemory hierarchy and scheduling reading IDue 4 October: Yu-Kwong Kwok and Ishfaq Ahmad.Benchmarking and comparison of the task graph schedulingalgorithms. J. of Parallel and Distributed Computing,59(3):381–422, 1999.Kaushik Ghosh, Bodhisattwa Mukherjee, and Karsten Schwan. Asurvey of real-time operating systems. Technical report, College ofComputing, Georgia Institute of Technology, February 1994Due 6 October: L. Yang, Robert P. Dick, Haris Lekatsas, andSrimat Chakradhar. High-performance operating systemcontrolled on-line memory compression. ACM Trans. EmbeddedComputing Systems, 9(4):30:1–30:28, March 2010.Due 11 October: .45Robert DickEmbedded System Design and SynthesisData compressionEmbedded system memory organizationOverview of real-time and embedded operating systemsEmbedded application/OS time, power, and energy estimationHomeworkUpcoming topicsTechnology trends.Power analysis and optimization.Emerging applications: CPS.Human-centered computer design.Energy supply in embedded systems.48Robert DickEmbedded System Design and Synthesis47Robert DickEmbedded System Design and Synthesis
Real-Time Operating Systems are often used in embedded systems They simplify use of hardware, ease management of multiple tasks, and adhere to real-time constraints Power is important in many embedded systems with RTOSs . Specialized or static memory management common 14 Robert Dick Embedded System Design and Synthesis
2. Embedded systems Vs General Computing system Page 4 Sec 1.2 ; 3. History of embedded systems , classification of embedded system Page 5,6 Sec 1.3 , Sec 1,4 . 4. Major application area of embedded sys Page 7 Sec 1.5 5. Purpose of embeded system Page 8 Sec 1.6 6. Typical Embedded sys: Core of embedded system Page 15 Chap 2 : 7. Memory Page 28
21-07-2017 2 Chap. 12 Memory Organization Memory Organization 12-5 12-1 Memory Hierarchy Memory hierarchy in a computer system Main Memory: memory unit that communicates directly with the CPU (RAM) Auxiliary Memory: device that provide backup storage (Disk Drives) Cache Memory: special very-high-
The network embedded system is a fast growing area in an embedded system application. The embedded web server is such a system where all embedded device are connected to a web server and can be accessed and controlled by any web browser. Examples; a home security system is an example of a LAN networked embedded system .
In memory of Paul Laliberte In memory of Raymond Proulx In memory of Robert G. Jones In memory of Jim Walsh In memory of Jay Kronan In memory of Beth Ann Findlen In memory of Richard L. Small, Jr. In memory of Amalia Phillips In honor of Volunteers (9) In honor of Andrew Dowgiert In memory of
Memory Management Ideally programmers want memory that is o large o fast o non volatile o and cheap Memory hierarchy o small amount of fast, expensive memory -cache o some medium-speed, medium price main memory o gigabytes of slow, cheap disk storage Memory management tasks o Allocate and de-allocate memory for processes o Keep track of used memory and by whom
Memory -- Chapter 6 2 virtual memory, memory segmentation, paging and address translation. Introduction Memory lies at the heart of the stored-program computer (Von Neumann model) . In previous chapters, we studied the ways in which memory is accessed by various ISAs. In this chapter, we focus on memory organization or memory hierarchy systems.
Chapter 2 Memory Hierarchy Design 2 Introduction Goal: unlimited amount of memory with low latency Fast memory technology is more expensive per bit than slower memory –Use principle of locality (spatial and temporal) Solution: organize memory system into a hierarchy –Entire addressable memory space available in largest, slowest memory –Incrementally smaller and faster memories, each .
counseling and consultation for little or no cost to the employee. VA offers up to 15 days a year of military leave support for reservists and National Guard, and supports our nurses’ ability to serve both their country and Veterans. VA employees have the benefit of the Federal Employee Retirement System and Thrift Savings Plan. VA also offers continuation of federal service from .
