Real-Time Systems. Design Principles for Distributed Embedded Applications. Herman Kopetz. Second Edition (811374), страница 57
Текст из файла (страница 57)
Which subsystem controls thespeed of communication if an explicit flow control schema is deployed?7.4 Calculate the latency jitter of a high level PAR protocol that allows threeretries, assuming that the lower level protocol used for this implementationhas a dmin of 2 ms and a dmax of 20 ms. Calculate the error detection latencyat the sender.Review Questions and Problems1897.5 Compare the efficiency of event-triggered and time-triggered communicationprotocols at low load and at peak load.7.6 What mechanisms can lead to thrashing? How should a system react ifthrashing is observed?7.7 Given a bandwidth of 500 Mbits/s, a channel length of 100 m and a messagelength of 80 bits, what is the limit of the protocol efficiency that can beachieved at the media access level of a bus system?7.8 How do the nodes in a CAN system decide which node is allowed to accessthe bus?7.9 Explain the role of the time-outs in the ARINC 629 protocol.
Is it possible fora collision to occur on an ARINC 629 bus?7.10 Describe the different types of time-controlled circuit switching!7.11 What services are provided by the TTP protocol?7.12 What is the self-confidence principle?7.13 Explain the principle of operation of time-triggered Ethernet!Chapter 8Power and Energy AwarenessOverview The increasing growth of energy-aware and power-aware computing isdriven by the following concerns:llllThe widespread use of mobile battery-powered devices, where the availabletime-for-use depends on the power consumption of the deviceThe power dissipation within a large system-on-chip that leads to high internaltemperatures and hot spots that have a negative impact on the chip’s reliability,possibly physically destroying the chipThe high cost of the energy for the operation and cooling of large data centers,and finallyThe general concern about the carbon emissions of the ICT industry, which is ofabout the same magnitude as the carbon emissions of the air-transport industry.In the past, the number of instructions executed by a computer system in a unit oftime (e.g., MIPS or FLOPS) was the important indicator for measuring the performance of a system.
In the future, the number of instructions executed per unit ofenergy (e.g., a joule) will be of equal importance.It is the objective of this chapter to establish a framework for developing anunderstanding for energy-efficient embedded computing. In the first section weintroduce simple models to estimate the energy consumption of different tasks. Thisgives the reader an indication of where energy is dissipated and what are themechanisms of energy dissipation. Since energy consumption depends very muchon the considered technology, we assume a hypothetical 100 nm CMOS VLSItechnology as the reference for the estimation.
The next section focuses on hardware techniques for energy saving, followed by a discussion about the impact ofsystem architecture decisions on the energy consumption. Software techniques thathelp save energy are treated in the following section. In the final section we look atthe energy content and management of batteries and the topic of energy harvestingfrom the environment.H. Kopetz, Real-Time Systems: Design Principles for Distributed Embedded Applications,Real-Time Systems Series, DOI 10.1007/978-1-4419-8237-7_8,# Springer Science+Business Media, LLC 20111911928.18 Power and Energy AwarenessPower and EnergyThe ICT sector is a major consumer of electric energy.
At the time of this writing,the carbon footprint of the ICT sector is in the same order of magnitude as thecarbon footprint of the air-transportation industry and growing at a rate of 10%per year.Example: Out of 156 of GW of electric power used for ICT equipment worldwide (about8% of the worldwide total electricity consumption) in the year 2007, about 28 GW (18%)are used for the operation of PCs, 22 GW (14%) are used for network equipment, 26 GW(17%) are used for data centers, 40 GW (25.5%) are used for TV, and 40 GW (25.5%) areused for other ICT purposes [Her09, p. 162].8.1.1Basic ConceptsThe concept of energy, initially introduced in the field of mechanics, refers to ascalar quantity that describes the capability of work that can be performed by asystem.
The intensity of the work, that is the energy used up per unit of time, is calledpower. Energy is thus the integral of power over time. There exist different forms ofenergy, such as potential energy, kinetic energy, thermal energy (heat), electricalenergy, chemical energy, and radiation energy.The first law of thermodynamics, relating to the conservation of energy, statesthat in a closed system, the total sum of all forms of energy is constant. Thetransformation of one form of energy to another form of energy is governed bythe second law of thermodynamics that states that thermal energy can only bepartially transformed to other forms of energy.
For example, it is impossible toconvert thermal energy to electrical energy with an efficiency of 100%, while theconverse, the transformation of electrical energy to thermal energy can be performed with 100% efficiency.There exist many different units to measure the amount of energy. We will usethe joule, which is the unit of energy in the mks (meter-kg-second) system.
Onejoule (J) is defined as the amount of work done by a force of one newton moving amass of one kg for a distance of one meter. At the same time a joule is the amount ofenergy that is dissipated by the power of one electric watt lasting for one second.Thermal energy, i.e., heat, is often measured in calories or BTUs (British ThermalUnits). One calorie is defined as the heat needed to increase the temperature of onegram of water at room temperature by one degree celsius.
One calorie correspondsto about 4.184 joule, while one BTU corresponds to about 1,055 joule.Example: A gram of gasoline contains chemical energy in the amount of about 44 kJ. Anautomotive engine converts about one third to one fourth of this chemical energy tomechanical energy, i.e., gram of gasoline provides about 12 kJ of mechanical energy.The rest of the chemical energy is converted to heat. The mechanical energy of 12 kJ issufficient to lift a car with a mass of 1,000 kg (or 1 t) about 1.2 m (the potential energy is8.1 Power and Energy193mgh, where g is the gravitational acceleration of 9.8 m/s2 and h the relative altitude inmeters) or to accelerate the car from zero m/s to a speed of 5 m/s (or 18 km/h), consideringthat the kinetic energy of the car is mv2/2.
If we stop the car by braking, this kinetic energyis converted into heat by the brakes and tires. In an electric car, a substantial part of thiskinetic energy can be transformed back into electrical energy that can be stored in a batteryfor further use.A battery is a storage device for electric energy. The voltage that exists at theterminals of a battery can drive electric current through a resistor.
In the resistor,electric energy is converted to heat. If an electric current I flows through a wire witha resistance R, the voltage drop on the wire will be, according to Ohm’s lawU ¼ IRand the dissipated electric power is given byW ¼ UIwhere W denotes the dissipated electric power (in watt), I the current (in ampere),and R the resistance (in ohm). If a constant voltage of U is applied to a system with aresistance of R over a time of t s, the energy that is dissipated equalsE ¼ tU2 =RWhere E is the dissipated energy in joule, t denotes the time in seconds and U and Rdenote the voltage and the resistance, respectively.8.1.2Energy EstimationThe energy that is needed by a computing device to execute a program can beexpressed as the sum of the following four terms:Etotal ¼ Ecomp þ Emem þ Ecomm þ EIOwhere Etotal is the total energy needed, Ecomp denotes the energy needed to performthe computations in a CMOS VLSI circuit, Emem denotes the energy consumed bythe memory subsystem, Ecomm denotes the energy required for the communication,and EIO is the energy consumed by I/O devices, such as a screen.
In the following,we investigate each of these terms in some detail by presenting simplified energymodels. The numerical values of the parameters depend strongly on the technologythat is used. We give some approximate value ranges for the parameters for ahypothetical 100 nm CMOS technology which is operated with a supply voltage of1 V in order to enable the reader to develop an understanding for the relativeimportance of each one of the terms.1948 Power and Energy Awarenessresistor RFig.
8.1 First-order RCnetworkvoltage Vcapacitance CComputation Energy. The energy Ecomp required for the computations consists oftwo parts: (1) the dynamic energy Edynamic that denotes the energy needed toperform the switching functions in a CMOS VLSI circuit and (2) the static energyEstatic that is dissipated due to a leakage current drawn from the power supplyirrespective of any switching activity.The dynamic energy consumption of a switching action of a transistor can bemodeled by a first-order RC network as shown in Fig. 8.1. In such a network,consisting of a power supply with voltage V (measured in volt), a capacitor C(measured in farad), and a resistor R (measured in ohm), the voltage on the resistorR as a function of time t (measured in seconds) after the switching event, is given byVres ðtÞ ¼ Vet=tWhere t ¼ RC, which has the dimension of second, is called the time-constant ofthe RC network.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
