You are on page 1of 6

A-Level Computer Science

Processor Types
Introduction

The architecture of CPUs and the way in which they work has changed over the years in
order to improve efficiency and effectiveness. The following pages will look at the
different types of processors, their features and effectiveness in a range of different
contexts. In particular we will look at the differences between:

Von Neumann, Harvard and Contemporary Architectures

Multicore and Parallel Systems

CISC and RISC

…and also look at the roles of GPUs.

Von Neumann, Harvard and Contemporary Architectures

In the previous lessons, all discussion surrounding the CPU related to a particular
architecture – ‘The Von Neumann Architecture’. This CPU design is used in the majority of
computers used today. As we have seen, with this architecture, programs instructions and
data are stored together in memory and each instruction and data item uses the same
data bus.

However, some computers, mostly specialist and embedded systems, use a different
architecture known as the ‘Harvard Architecture’. With this architecture, a program’s
instructions and data are stored separately in memory with the instructions and data item
using separate data buses.

Computer Science UK Membership Site Licence: Do not share outside your centre. PAG
E \*
A-Level Computer Science

Von Neumann VS Harvard Architectures

Contemporary Architectures

In modern day processors, the Von Neumann architecture is more widely used. It is after
all cheaper and makes more efficient use of the RAM as both data and instructions can
be stored together.

However, most modern processors will make use of the Harvard architecture in their
cache, dividing the cache memory into separate parts, separating data and instructions
taking advantage of the benefits of parallel processing.

CISC and RISC CPUs

As we have seen before, the CPUs control units will contain an instruction set. An
instruction set is a list of instructions that the control unit can understand and can perform
on the data.

Different CPUs may however use one of two types of instruction set:

CISC – Complex Instruction Set Computers

RISC – Reduced Instruction Set Computers

Computer Science UK Membership Site Licence: Do not share outside your centre. PAG
E \*
A-Level Computer Science

The following will discuss each, looking at their relative advantages and disadvantages.

CISC Processors

This type of instruction set will have a large amount of known instructions (for example 80-
100) and these will be complex in nature (i.e. each instruction may represent several
steps).

For example, there may be an instruction to multiply 2 numbers. In CISC processors, the
actual hardware will be built to automate this instruction and so will LOAD the values,
MULTIPLY them and STORE the result in one operation (which may take several clock
cycles).

Advantages of this are:

Because a complex instruction (e.g multiply) is known to the CPU, it is easy for a compiler
to convert a high level language multiply statement into the appropriate assembly
language instruction.

Because it is just one instruction (which may represent


several steps), it requires little memory to be stored.

RISC Processors

This type of instruction set has a much smaller amount of known instructions (for example
30-40). For example, although there may also be an instruction to multiply 2 numbers, in a
RISC CPU, this instruction will only represent the multiplication part of the process and not
the LOADING and STORING of values. In RISC processors, the instructions are not complex
(they are simple) and so a single RISC instruction will not automate all other required steps
such as LOAD and STORE. As a result, in RISC, more instructions are required to perform a
task.

Although it may seem that RISC is more limited, it actually has a major advantage over
CISC. Because each RISC instruction is simple, each one takes a single clock cycle to be
executed. This means that RISC processors can benefit from pipelining, whereby as one

Computer Science UK Membership Site Licence: Do not share outside your centre. PAG
E \*
A-Level Computer Science

instruction is being executed, the next can be decoded and the next fetched all at the
same time. CISC instructions would have to be fully executed before the next could be
fetched. As a result, RISC instructions will execute at least as fast as CISC.

CISC vs RISC Processors

Multicore and Parallel Systems

Computer Science UK Membership Site Licence: Do not share outside your centre. PAG
E \*
A-Level Computer Science

Multi core processors have more than one processing unit and so ‘in theory’ can carry out
multiple jobs at once. This principal of carrying out multiple jobs at once is known as
parallel processing and its effects can be that more processes execute over the same
time period. However, a major issue of this is that software has to be written to specifically
allow multiple jobs to be carried out simultaneously. So, as many older pieces of software
are not written to make use of this technology, multi-core and parallel systems do have
limitations.

Graphics Processing Units (GPUs)

Continuing on from the notion of parallel processing, GPUs are particularly good at
processing multiple jobs in parallel (at the same time). A GPU is a ‘Graphics Processing
Unit’, which traditionally is responsible for the processing of large blocks of visual data,
very quickly. Its architecture consists of a great deal of cores, enabling it to handle
multiple processes at once. Often GPUs are separate cards
added to a system but they can also be embedded on
motherboards. Either way, they are known as co-processors as
they provide extra processing support for the CPU. They do this
taking on some of the computer intensive jobs freeing up more
capacity for the CPU.

Whereas a multi-core CPU can carry various instructions on different data, in parallel, a
GPU is designed to carry out one instruction on multiple pieces of data in parallel.

As a result, software engineers have taken advantage of this fact and now, many other
(non-graphic related) jobs are taken on by the GPU.

For example:

● Machine learning (AI)

● Modelling

● Cryptocurrency Mining (e.g. mining for bitcoins)

Computer Science UK Membership Site Licence: Do not share outside your centre. PAG
E \*
A-Level Computer Science

Questions

‘General Understanding’ Questions

1. Explain how CISC and RISC processors work, comparing their relative advantages
and disadvantages. [6]

2. Explain (using examples) why GPUs are known as co-processors. [4]

3. Explain the differences between Von Neumann and Harvard CPU architectures. [6]

4. Explain how both Von Neumann and Harvard architectures are used in modern
CPUs. [3]

Computer Science UK Membership Site Licence: Do not share outside your centre. PAG
E \*

You might also like