Types of Processors

Question 1

Instruction Set

Answer 1

The list of all instructions the processor understands

Question 2

CISC

Answer 2

Complex Instruction Set Computing

If you want to multiply two numbers x * y
there is an instruction set which will do the multiplication
- MULT x, y

When the complier sees a multiplication in the code it will convert it into a single machine code instruction

However to complete this instruction it might take multiple clock cycles, this requires complex circuitry

So trade off complex hardware but simpler coding

Question 3

Examples of CISC

Answer 3

Desktop and laptop computers

Question 4

RISC

Answer 4

Reduced Instruction Set Computing

If you want to multiply two numbers x * y, in RISC the complier will need to generate multiple lines of low-level machine code
- LOAD x from memory register 1
- LOAD y from memory register 2
- PROD register1, register2
- STORE into memory location

this is an increase of instructions from 1 to 4 compared to CISC
we can use pipelining to help increase efficiency
means we can have simpler hardware

Question 5

Example of RISC

Answer 5

used in mobile phones

Question 6

CISC VS RISC (there’s 7)

Answer 6

CISC is used in laptops and desktops computers, made by Intel or AMD
RISC is used in smartphone and tablets, based around ARM processor

CISC has more complex hardware
RISC has simpler hardware

CISC has multiple machine cycles per instructions
RISC has a single machine cycle per instruction

CISC has physically larger in size and require more silicon to make thus more expensive
RISC is smaller in size as less complex circuitry required, less silicon needed to make thus cheaper

CISC has greater energy consumption
RISC has lower energy requirements, and can go into “sleep mode” when not actively processing

More intensive tasks will do better with CISC
RISC runs at a lower clock speed, but can perform simpler tasks more quickly than CISC

CISC cannot support pipelining
RISC can support pipelining

Question 7

Co processors

Answer 7

A co-processor is any other processor used for a specific task e.g.
I/O interfacing or encryption, string processing, floating-point arithmetic
they help to improve the performance of the computer but carrying out the instructions in parallel with the main CPU

Question 8

GPUs

Answer 8

Graphics Processing Units

used for rendering Graphics
they have a large number of cores
this means they can process many streams of parallel data at the same time
each of the GPU processor streams (100s or 1000s) run at a lower speed than a CPU

Question 9

GPU and CPU

Answer 9

CPUs - Good at complex operation on small data sets
GPUs - Good at simple operations on large data sets

GPUs are like a special type of CPU that can execute a single instruction over a large amount of data in parallel

Question 10

GPU uses

Answer 10

Artificial intelligence/deep learning systems
breaking encryption schemes
scientific modelling of physical systems, such as modelling the formation of a hurricane
real time signal processing, such as handling satellite imagery

Question 11

Sequential Processing - what is it

Answer 11

Sequential processing is when instructions are carried out, one at a time, in the order in which they were received

Question 11

Sequential Processing - how it works

Answer 11

A sequential processor would examine each pixel one at a time and apply the processing instruction

it cannot change every pixel at the same time

Question 12

Multicore systems

Answer 12

Multicore - A single chip (Chip Multiprocessor CMP) with multiple processing units on it; cores (Dual, Quad)

Each core can execute CPU instructions on their own: ADD, STORE, LOAD

Cores may share a local cache or inter-core communication

Doubling the number of cores DOES NOT MEAN double the performance

Question 13

Parallel Processing systems

Answer 13

Working alongside multiple cores

the processing of program instructions by dividing them amongst multiple processors / processor cores

Question 14

Parallell Processing example

Answer 14

In a normal processing, each core takes on a separate part of the FDE cycle

Parallel processing with pipelines, a single job is split into tasks. Main processing fetches and decodes the instruction and farms off parts to be executed

Question 15

Types of Parallel Processing - SIMD

Answer 15

An array processor has several ALUs. this allows several elements of an array to be processed at the same time rather than in a sequence

Single Instruction Multiple Data (SIMD)
A single instruction is issued by a control unit and that instruction is applied to a number of data sets at the same time

Games consoles and graphics cards make heavy use of array processors to shift those pixels about

-ve = This architecture relies on the fact that the data sets are all acting on a single instruction. However, if these data sets somehow rely on each other than you cannot apply parallel processing

Question 16

Types of Parallel Processing - MIMD

Answer 16

The next level of parallel processing is to have multiple instructions acting upon multiple data sets. this is called MIMD or Multiple Instruction Multiple Data

Focuses the efforts of multiple CPUs to a single task, with each CPU carrying out only part of the overall task. A program called ‘Job Scheduler’ is responsible for assigning these parts to each core

-ve = Dependent on task being able to cut a problem down into chunks that can be processed independently. But not many problems can be broken down this way

Programmers also have to write specific code to make use of the multi-core CPU. This is actually quite difficult and even now most applications running on a multi-core CPU will not be making full use of all the cores

Question 17

Advantages of Parallel Processing

Answer 17

• Faster than one processor when handling lots of data, with each component of data requiring the same processing
• Not limited by bus transfer rate
• Can make maximum use of CPU despite bottlenecks

Question 18

Disadvantages of Parallel Processing

Answer 18

• Only certain types of problems are suitable, a problem where its component parts depend on each other cannot be parallel processed
• More costly as more processing blocks are needed
• Software is harder to write and a complex OS is required to communicate with multiple processors