Core of emb-sys

7/31/2019 Core of emb-sys

1/52

Core Of The Embedded System


2/52

Embedded systems are domain and

application specific and are build around a

central core.


3/52

Categories

The core of embedded system falls into any

one of the following categories:

General purpose and domain specific processors

Microprocessors

Microcontrollers

Digital Signal Processors

Application Specific Integrated Circuits(ASICs)

Programmable Logic Devices(PLDs)

Commercial off-the-shelf Components(COTS)


4/52

General Purpose & Domain Specific Processors

Almost 80% of Embedded Systems areprocessor/Controller based.

The processor may be a micro-processor or a

microcontroller or a digital signal processordepending on the domain and application.

Eg. Embedded systems in industrial control andmonitoring applications make use of microprocessor

or microcontroller where as domains which requiresignal processing such as speech coding, speechrecognition make use of digital signal processors.


5/52

Microprocessors

A microprocessor is a silicon chip representing aCPU capable of performing arithmetic as welllogical operations according to a predefined setof instructions, which is specific to the

manufacturer. It contains Arithmetic and logic unit, control unit

and working registers.

A microprocessor is a dependant unit and it

requires combination of other hardware likememory, timer unit, and interrupt controllerect.For proper functioning.


6/52

Microprocessors

Different instruction set and systemarchitecture are available for the design of amicroprocessor.

Harvard and von-neumann are two commonarchitectures for processor design.

Processors based on harvard architecture containsseprate buses for program memory and datamemory, whereas processors based on Von-Neumann architecture shares a single system busfor program and data memory


7/52

Microprocessors

Reduced Instruction Set Computing(RISC) and

Complex Instruction Set Computing(CISC)are

two common Instruction Set Architectures

(ISA) available for processor design.


8/52

General Purpose Procesor (GPP)vs. Application

Specific Instruction Set Processor

A GPP is designed for general computationaltasks. They are produced in large volumes andper unit cost for a chip is low compared to

ASIC or other Ics. On other hand ASIPs have architecture and

instruction set optimized to specificdomain/application requirement like networkprocessing, automotive, telecom, digital signalprocessing and control applications.


9/52

Microcontrollers

A microcontroller is a highly integrated chip

that contains a CPU, a scratch pad RAM,

special & general purpose register arrays, on

chip ROM/FLASH memory for program

storage, timer and interrupt control units and

dedicated I/O ports.

It can be considered as a super set ofmicriprocessor.


10/52

Microcontrollers

Microcontrollers found greater place in

embedded domain since they contain all

necessary functional blocks for independent

working.

They are cheap, cost effective and are readily

available in market.


11/52


12/52

Microprocessor vs. Microcontroller

Microprocessor

A silicon chip representing acentral processing unit, which iscapable of performing arithmetic

as well as logical operationsaccording to a predefined set ofinstructions.

It is a dependant unit. It requiresthe combination of other chipslike timers, program & data

memory chips, interruptcontrollers etc for functioning.

Most of the time general purposein design and operation.

Microcontroller

It is a highly integrated chip thatcontains a CPU, scratchpad RAM,special & general purpose register

arrays, on chip ROM, Flashmemory for program storage,timer & interrupt control unitsand dedicated I/O ports.

It is a self contained unit & it doesnot require external interrupt

controller, timer, UART etc for itsfunctioning.

Mostly application-oriented ordomain specific.


13/52

Microprocessor vs. Microcontroller

Microprocessor

Doesnt contain a built in I/O

port. The I/O port functionality

needs to be implemented with

the help of a external

programmable peripheral

interface chip like 8255.

Targeted for high end market

where performance isimportant.

Limited power saving options

compared to microcontrollers.

Microcontroller

Most of the processors contain

multiple built-in I/O ports

which can be operated as a

single 8 or 16 0r 32 bit port or

as individual port pins.

Targeted for embedded

market where performance is

not so critical(At present thisdemarcation is not valid).

Includes lot of power saving

features.


14/52


15/52

Digital Signal Processors

DSPs implement algorithms in hardware whichspeeds up the execution where as generalpurpose processors implement the algorithm

in firmware and speed of execution dependson clock for the processors.

In general DSP can be viewed as a micrichipdesigned for performing high speedcomputational operations for addition,subtraction, multiplication & division.


16/52

Key Units of a DSP

Program memory Memory for storing the program required by DSP to

process the data.

Data Memory

Working memory for storing temporary variables anddata/signal to be processed.

Computational Engine Performs the signal processing in accordance with the

stored program memory. Computational engine

incorporates many specialized arithmetic units and each ofthem operates simultaneously to increase the executionspeed. It also incorporates multiple hardware shifters forshifting operands and thereby saves execution time.


17/52


18/52

RISC vs. CISC Processors/Controllers

The term RISC stands for Reduced Instruction

Set computing.

All RISC processors/controllers posses lesser

number of instructions in the range of 30 to

40.

CISC stands for Complex Instruction Set

Computing. Its instruction set is complex and

instructions are high in number.


19/52

RISC vs. CISC Processors/Controllers

From a programmer point of view RISC

processors are comfortable because he/she

needs to learn only few instructions where as

in CISC processor more instructions arerequired to be learnt and should understand

the context of usage of each instruction.


20/52

RISC vs CISC

RISC

Lesser number of instructions.

Instruction pipelining andincreased execution speed.

Orthogonal instruction set (allows each instruction tooperate on any register anduse any addressing mode).

Operations are performed onregisters only, the onlymemory operations are loadand store.

CISC

Greater no of instructions.

No instruction pipeliningfeature.

Non-orthogonal instruction set(all instructions are notallowed to operate on anyregister and use anyaddressingmode. It is instruction

specific).

Operations are performed onregisters or memorydepending on the instruction.


21/52

RISC vs. CISC

RISC

A large number of registers areavaiable.

Programmer needs to write

more code to execute a tasksince the istructions aresimpler ones.

Single, fixed lengthinstructions.

Less silicon usage and pincount.

With havard architecture.

CISC Limited number of general purpose

registers.

Instructions are like macros in clanguage.

A programmer can acheive the desiredfunctionality with a single instructionwhich in turn provides the effect ofusing more simpler single instructionsin RISC.

Variable length instructions

More silicon usage since more

additional decoder logic is required toimplement the complex instructiondecoding.

Can be harvard or von-neumanarchitecture.


22/52

Harvard vs. Von-Neuman

Processor/Controller Architecture The terms harvard and von-neuman refers to the processor

architecture design.

Microprocessors/controllers based on the Von-Neumanarchitecture share a single common single bus for fetchingboth instructions and data.

Program instructions and data are stored in a commonmain memory.

Von-Neuman based processors/controllers first fetch aninstruction and then fetch the data to support the

instruction fron code memory which slows down thecontrollers operation.

It is also known as Princeton architecture, since it wasdeveloped by the Princeton University.


23/52

Harvard vs. Von-Neuman

Processor/Controller Architecture Microprocessors/controllers based on the Harvard

architecture will have separate data bus and instruction buswhich allows the data transfer and program fetching tooccur simultaneously on both buses.

With harvard architecture, the data memory can be readand written while the program memory is being accessed.

These separated data memory and code memory busesallow one instruction to execute while the next instructionis fetched (pre-fetching).

The pre fetching theoretically allows much faster execution thanvon-neuman architecture.

Since some additional hardware logic is required for thegeneration of control signals for this type of operation it addssilicon complexity to the system.


24/52

Harvard vs. Von-Neuman Architecture

I/O CPU Memory

ProgramMemory

CPU Data MemorySingle Shared Bus


25/52

Harvard vs. Von-Neuman Architecture

Harvard Architecture

Separate buses for instructionand data fetching.

Easier to pipeline, so high

performance can be achieved . Comparatively high cost.

No memory alignmentproblems.

Since data memory and

program memory are storedphysically in differentlocations, no chances foraccidental corruption ofprogram memory.

Von-Neuman Architechture

Single shared bus forinstruction and data fetching.

Low performance compared to

harvard architecture. Cheaper.

Allows self modifying codes.

Since data memory andprogram memory are stored

physically in the same chip,chances for accidentalcorruption of programmemory.


26/52

Big-Endian vs. Little Endian

Processors/Controllers

Endianness specifies the order in which the datais stored in the memory by processor operationsin a multibyte system(processors whose word sizeis greater than one byte). Suppose the word

length is two byte then data can be stored inmemory in two different ways: Higher order of data byte at the higher memory and

lower order of data byte at location just below thehigher memory.

Lower order of data byte at the higher memory andhigher order of data byte at location just below thehigher memory.


27/52

Little endian

Following figure means the lower order byteof

the data is stored in memory at the lower

address and higher order byte at the highest

address. (The little end comes first).Forexample, a 4 byte long integer Byte3 Byte2

Byte1 Byte0 will be stored in the memory as

shown below:


28/52

little-endian

Byte0

Byte1

Byte2

Byte3

Base Address+0 Byte 0


Base Address+2 Byte2


0x20000 (Base Address)

0x20001 (Base Address +1)

0X20002 (Base Address +2)

Ox20003 (Base Address +3)


29/52

Big-endian

Means the higher order byte of the data is

stored in memory at the lowest address and

the lower-order byte at the highest address.

(The big end comes first.) For example, a 4byte long integer Byte3 byte2 Byte1 Byte0 will

be stored in the memory as follows:


30/52

Big-endian

Byte3

Byte2

Byte1

Byte0



Base Address+2 Byte1


0x20000 (Base Address)

0x20001 (Base Address +1)

0X20002 (Base Address +2)

Ox20003 (Base Address +3)


31/52

Application Specific Integrated Circuit

Application Specific Integrated (ASIC) is amicrochip designed to perform a specific orunique application.

It is used as replacement to conventionalgeneral purpose logic chips.

It integrates several functions into a single

chip and thereby reduces the systemdevelopment cost.

Most of the ASICs are proprietary products.


32/52


As a single chip ASIC consumes a very small area in the totalsystem and thereby helps in design of smaller systems withhigh capabilities/functionalities.

ASIcs can be pre fabricated for a special application or itcan be custom fabricated by using the components from areusable building block library of components for aparticular customer application.

ASIC based systems are profitable only for large volumecommercial productions.

Fabrication of ASICs requires a non-refundable initialinvestment for the process technology and configurationexpenses. This investment is known as NRE and it is onetime investement.


33/52


If non-recurring engineering charges is borneby a third party and the application SpecificIntegrated Circuit(ASIC) is made openly

available in the market, the ASIC is referred asApplication Specific Standard Product (ASSP).

The ASSP is marketed to multiple customersjust as a general purpose product is, but to a

smaller number of customers since it is for aspecific application.


34/52


Since ASICs are proprietary products, thedevelopers of such chips may not beinterested in revealing the internal details of it

and hence it is very difficult to point out anexample of it.

Moreover it will create legal disputes if anillustration of such an ASIC product is given

without getting prior permission from themanufacturer of the ASIC.


35/52

Programmable Logic Devices

Logic devices provide specific functions, including

device to device interfacing, data communication,

signal processing, data display, timing and control

operations, and almost every other function asystem must perform.

Logic devices can be classified into two broad

categories: Fixed

Programmable


36/52


As the name indicates, the circuit in a fixed logicdevice are permanent, they perform one functionor set of functionsonce manufactured, theycannot be changed.

On the other hand programmable logicdevices(PLDs)offer customer a wide range of logiccapacity, features, speed, and voltage

characteristics- and these devices can bereconfigured to perform any number of functionsat any time.


37/52


With programmable logic devices, designers useinexpensive software tools to quickly develop,simulate and test their design. Then a design canbe quickly programmed into a device,immediately tested in a live circuit.

The PLD that is used for this programming is exactsame PLD that will be used in the final production

of a piece of end equipment, such as a networkrouter, DSL modem, a DVD player or anautomotive navigation system.


38/52


ryThere are no NRE costs and the final design is completedmuch faster than that of a custom, fixed logic device.

Another key benefit of using PLDs is that during the designphase customers can change the circuitary as often as theywant until the design operates to their satisfaction.

It is because PLDs are based on re-writeable memorytechnology- to change the design, the device is simplyrepeogrammed.

Once the design is final the customers can go into

immediate production by simply programming as manyPLDs as they need with final software design file.


39/52

CPLDs and FPGAs

Programmable logic devices are of two types Field Programmable Gate Arrays(FPGAs)

Cmplex Programmable Logic Devices(CPLDs)

FPGAs offer highest amount of logic density.

The largest FPGGA provides 8 million system gates. They also have built-in hardwired processors,

substantial amount of memory, clock managementsystems,and support for lates very fast device to device

signaling technologies. FPGAs are used in applications as data processing,

instrumentation, telecommunication and digital signalprocessing.


40/52

CPLDs and FPGAs

CPLDs offer much smaller amounts of logic-

about 10,000 gates

They offer very predictable timing characteristics

and are therefore ideal for critical controlapplications.

They require low amounts of power and are very

inexpensive making them ideal for cost sensitive,battery operated, portable applications such as

mobile phones and digital hand held assistants.


41/52

Advantages of PLDs

Programmable logic devices offer number ofadvantages over fixed logic devices: They offer much more flexibility during the design cycle

because design iterations are simply a matter of changingthe programming file and the result of design changes canbe seen immediately in working parts.

They do not require long lead time for prototypes orproduction parts-the PLDs are already on a distributorsshelf ready for shipment.

They do not require customer to pay for large NRE costsand purchase expensive mask setsPLd supplier incurrthose costs when they design their programmable devicesand are able to amortize those costs over the multi-yearlifespan of a given line of PLDs.


42/52

Advantages of PLDs

PLDs allow customers to order just the number of

parts they need, when they need them, allowing

them to control inventory. Customers who use

fixed logic devices often end up with excessinventory which must be scrapped or if demand

for their producr surges, they may be caught

short of parts and face production delays.

PLDs can be reprogramme even after a piece of

equipment is shipped to a customer.


43/52

Commercial Off-the-Shelf Components

(COTS)

A commercial off-the-shelf product is thatwhich is used as-is.

COTS products are designed in such a way to

provide easy integration and interoperabilitywith existing system components.

The COTS component itself may be developedaround a general purpose or domain specificprocessor or an application specific Integratedcircuit or a programmable logic device.


44/52


(COTS)

Typical examples of COTS hardware units are

remote controlled toy car control units

including the RF circuitory part, high

performance, high frequency microwaveelectronics (2-200GHz), high bandwidth

analog to digital convertor, devices and

components for operation at very hightemperatures, electro-optic IR imaging arrays

UV/IR detectors etc.

ff f


45/52


(COTS)

The major advantage of using COTS is that they

are readily available in the market, are cheap and

a developer can cut down his development time

to a great extent, which in turn reduces the timeto market embedded systems.

Though multiple vendors supply COTS for the

same application, the major problems faced byend user is that there are no operational and

manufacturing standards.

l ff h h lf


46/52


(COTS)

A commercial off the shelf component

manufactured by a vendor need not have

hardware plug-in and firmware interface

compatibility with one manufactured by asecond vendor for the same application.

This restricts the end user to stick to a

particular vendor for a particular COTS. Thisgreatly affects the product design.

i l ff h h lf


47/52


(COTS)

The major drawback of using COTS

components in embedded design is that the

manufacturer of the COTS component may

withdraw the product or discontinue theproduction of COTS at any time if a rapid

change in technology occurs and this will

adversely affect a commercial manufacturer ofthe embedded system which makes use of the

specific COTS product.


48/52


49/52


50/52


51/52


52/52

Core of emb-sys

Documents