CHAPTER 3 - HARDWARE

The Von Neumann Concept

The Von Neuman Concept is an important one as it means that computers no
longer have to look like this...

Before 1943 computers like the one above did not store their programs in memory,
instead a series of switches had to be manually turned, or read from a paper
stream. These instructions would then be read one at a time. If you required the
computer to do something else you would then have to manually turn all pf the
switches into another combination.

This situation was far from ideal, time consuming and prone to mistakes.

In 1943 a mathematician named Von Neuman came up with the idea of storing these
program instructions and the data used into the same memory.
This is the basis of the Von Neuman concept - That program instructions and data
can be stored on the same memory.

Fetch / Decode / Execute

With program instructions and data now being stored in the same memory, a
process had to be designed to allow CPUs (Central Processing Units) to find the
correct instruction and execute it... after all its no longer as simple as reading the
switches!.

The process that was designed is known as the FETCH - EXECUTE - CYCLE

These three operations are constantly repeated by the CPU as the computer is
running this allows the CPU to quickly and efficiently work through new instructions
as they are required.

The Three Stages

 Fetch
This operation is triggered by
the CPU causing the next
instruction AND data to be
'fetched' from memory
(RAM).

Decode
Once the CPU has fetched the
instructions and data it will
then translate them into
instructions understandable by
the CPU

Execute
The CPU will then execute
the instructions in a logical
sequence

Once the execute operation is

complete, the CPU will repeat
the process by returning to the
'fetch' operation.

The Central Processing Unit (CPU)

The CPU is commonly referred to as the brain of the computer and this is a fair
comparison. THE CPU is responsible for all processing and calculations within a
computer.
The above diagram should illustrate the fact that the CPU sits firmly at the center of
a computer connecting and managing all components as they interact with each
other.

CPU Internal Components

The CPU itself is actually a small 45mm X
42.5mm chip but don't be fooled by its size.
Within this small squared space there is
actually a lot going on.

Within the CPU you will find 4 main

components, these are:

• Control Unit

• Arithmetic and Logic Unit (ALU)

• General Purpose Registers

• Program Counter

Control Unit Arithmetic and

The Control unit controls and coordinates the operation of Logic Unit (ALU)
the computer by controlling which instructions are The ALU is responsible for all
executed and when. mathematics and calculations
performed by the CPU. It can only
The CU contains the Current instruction register perform the following calculations:
(CIR) and the Program counter (PC)

Addition
The CIR contains the instruction currently being executed, Subtraction
the PC contains the address if the next instruction. Multiplication
Division
The CU will fetch an instruction from main memory Logical Comparisons e.g. Is A =
(address found in the PC)and then pass it on to the other B, Is 3 > 5
CPU parts to execute it effectively.

The ALU can also increase

complexity of these logical
 comparisons by including boolean
operators AND, OR & NOT

E.g. Is A < B OR C <A

Program Counter General Purpose

The Program Counter is used to store the address in Register
memory of the next instruction to be executed. The GPR can be thought of as the
notepad of the CPU. This is used to
hold data when performing
calculations. If the ALU needs to
temporarily store data as part of its
calculations, the GPU is where it is
stored.

Other Registers used

Within the CPU and its Memory unit there are a number of registers used. A register
is simply a small but very fast memory unit. The Program counter and General
purpose registers are examples of registers and their jobs have been described
above, however there are a few more important ones all of which are described
below.

Memory Address Memory Data

Register (MAR) Register (MDR)
The purpose of the MAR is to hold the location of
data that needs to be accessed. This is address is When the CPU looks into the memory
the physical location of data (instruction) that is address stored in the MAR it will find data
being stored in the computers memory. (An instruction) this instruction is then
 temporarily stored in the MDR before
being copied into the CIR.

REGISTER SUMMARY

All Registers used in the Von Neuman Architecture are:

PC (Program counter)
Contains the address of the next instruction to be fetched

MAR (Memory Address Register)

Contains the address of the current instruction to be fetched

MDR (Memory Data Register)

Contains the instruction / data after it is fetched from memory

CIR or IR ((Current) Instruction Register)

Contains the instruction / data to be decoded

ACC (Accumulator)
Contains the results of any calculations performed in the ALU

Coming Together
As you can see there are a lot of different parts within the CPU, all of which have to
operate under the Fetch - Decode - Execute Cycle concept. Luckily a system has
been developed where by all parts work in perfect harmony to maintain the
cycle. Follow the diagram below to fully understand the process in more detail.
 Buses
Within the computer system and within the CPU itself there are many different
components that have to work together. In order to communicate with each other
there needs to be some sort of connection between them that will allow for data
transmission.
This connection comes in the form of buses. A bus is a set of parallel wires that
connect two more components within a computer system.

The bus can be broken down into three different wires, these are:

Address bus - This carries signals that relate to addresses between the processor
and memory. It is Uni-directional which means that data will only travel in one
direction.

Control bus - This carries signals that relate to control e.g. an instruction to read
data. This can be uni-directional or Bi-directional.

Data Bus - This will carry actual data between components and devices. It is bi-
directional which means data can travel in both directions.

CPU CORES
In the CPU a core is essentially a processor unit that is capable of fetching, decoding and
executing instructions independently. In the past, CPUs would only contain one core,
however, in recent years CPU manufacturers have managed to create CPUs containing
between 2 and 32 cores.
What this means is that the modern CPU will contain between 2 and 32 processors... but why
would we want this?

Having multiple cores in a CPU makes it capable of completing multiple tasks at the same
time (Multitasking). This drastically increases the performance of the CPU as more tasks can
be complete in less time.

Each CPU core contains its own ALU, Control Unit and Registers thus allowing it to work on
tasks independently.

MORE CORES = MORE POWER?

It is true that adding more cores to a CPU will increase the overall performance. This
technology has allowed us to stop chasing higher clock speeds (covered in the next section).
It is important to realise that increasing the number of cores will not result in a direct 100%
performance improvement. This means that doubling the number of cores will not result in
exactly double the performance. Quadrupling the number of cores will not result in exactly 4
times the performance.
Why?

Well each time you add a new core to the CPU it increases the channels of communication
and the amount of coordination that needs to take place between them.

CLOCK SPEED
As you can see above, the CPU pictured has 8 - 16 cores. This detail is closely followed by a
"Clock speed" of 3.80 GHZ.

What is clock speed?

Clock speed essentially refers to the speed that your CPU cores can execute instructions. The
higher the clock speed, the faster your CPU can operate.

The CPU will process a large amount of low level instructions every second. The clock
speed, (generally) referred to as GHz (gigahertz) measures how many cycles your CPU can
execute per second.

The actual unit used to describe a processors speed will of course depend on how many
instruction cycles it can carry out per second. The units used are detailed in the table below:

Unit Cycles per second

1 Hertz (Hz) 1 instruction per second
1 Kilohertz (kHz) 1000 Cycles per second
1 Megahertz (mHz) 1,000,000 Cycles per second
1 Gigahertz (Ghz) 1,000,000,000 Cycles per second
1GHz = 1 billion CPU cycles per second!

This means that the core clock speed in our pictured CPU allows for 3.8 billion cycles per
second.

In General you can assume that the higher the clock speed the faster the performance of
the CPU will be.

CPU CACHE
Modern CPUs are fast, so fast in fact that their processing speeds can often out pace the speed
of RAM. This causes the CPU to slow down as it deals with the speed bottle neck.

To try and combat this a small but extremely fast piece of memory is built into the CPU
called the cache.
Cache Levels
There are three levels of cache within a CPU, each with differing speed size and
responsibility. The three levels are referred to as L1, L2 and L3 Cache.

L1 Cache - This is the fastest of all three caches and it stores the data that the CPU is most
likely to use whilst completing current tasks. It is very fast but also very small in size.
Typical L! caches will be between 2 and 64 KiB.

L2 Cache - This is slightly slower than the L1 cache but it does have a larger storage
capacity. L2 cache size can vary between 256KiB and 8MiB. Although the L2 cache is
slower than the L1 cache it is still approximately 25 times faster than RAM.

All cores in a CPU will have access to their own individual L1 and L2 caches. This is
illustrated in the diagram above.

L3 Cache - This cache is slower than the L1 and L2 caches, however it is still much faster
than RAM. L3 caches are shared between all CPU cores as illustrated in the above diagram.
The storage capacity of L3 cache is far superior to that of L1 and L2. Most decent CPUs
these days will offer L3 cache with a capacity of between 32 and 64 MiB.

What does the cache store?

The cache will store commonly used functions / data that are normally held in the
RAM. When a CPU needs to process data it will first look in the Cache. If the data is found
in the cache it is known as a cache hit! If the data is not found it is referred to as a cache
miss. After a cache miss the CPU must then look in the RAM for the data required.
To help further improve performance the CPU cache will attempt to predict what data will be
required in the cache For example, if a CPU is processing a video file, the cache may attempt
to retrieve additional parts of the file ready for process.

Typically each core within a CPU will have its own cache to store data that it will process.

The larger the CPU cache the better it can perform as more data can be quickly
accessed from the ultrafast memory.

CPU - INSTRUCTION SETS

INSTRUCTION SETS
In a CPU, the instruction set refers to the different types of operations that a CPU is
capable of performing.

When we talk about instruction sets we are referring to the low level machine code
instructions that a CPU is built to handle.
Low level operations are split into an opcode and an operand.

Opcode - Represents the machine instruction i.e. what it has to do - ADD, LOAD,
STORE or SUBTRACT

Operand - This is either the value to be operated on OR the address that holds the
value to be operated on.

Each CPU only has a limited number of opcodes available to them and the available
opcodes is what is referred to as the instruction set.

Interesting Point
The available operations are hard coded into each CPU and the consist of electronic
circuitry and transistors. Different instructions will allow the CPU to enable / switch
on relevant transistors to perform the required tasks.

INSTRUCTIONS
ach CPU has a limited number of instructions that it is capable of performing. As
the instructions are hard coded with transistors it is obviously not possible to add
more with a software update. Some example instructions include:

Operation Purpose
ADD Add two numbers
IN Allow input from a device
SUB Subtract one number from another
DIV Divide one number by another
STORE Store a value in a given location
MUL Multiply 2 numbers
HALT Stop operations