1.1.2 types of processor Flashcards
1
Q
What is an instruction set and what are the 2 types?
A
- Set of all instructions written in machine code that can be recognised and executed by a given processing unit/CPU
- CISC AND RISC
2
Q
What is RISC
A
- Reduced Instruction Set Computing
- Instruction set made up of a small number of simple, fixed-length instructions/commands
- 1 instruction executed per clock cycle (designed to process simple instructions quickly)
3
Q
RISC examples of manufacturers
A
- ARM processors
4
Q
Where are RISC processors suitable for use and why?
A
- tablets, smartphones, and embedded systems, where it is not necessary to carry out complex instructions.
5
Q
Advantages of RISC
A
- simpler instruction design (all instructions are uniform in terms of execution time) makes pipelining easier to implement –> increases performance of RISC.
- simpler instructions mean programs run faster
- require less circuitry (fewer transistors) –> cheaper to design and produce, give off less heat, consume less power and leaves more room for general purpose registers and cache
-lower energy requirements
6
Q
Disadvantages of RISC
A
- several instructions needed therefore low-level programs for RISC processors are longer and take up more memory.
- Uses a lot RAM to store assembly instructions (more lines of code required). Can cause system bottlenecks if RAM is limited
- Emphasises software over hardware: high-level compilers that simplify and translate programmers code do more work
- larger code sizes (lower cycles per second)
- fewer addressing modes available
7
Q
What is CISC
A
- Complex Instruction Set Computing
- large number of complex instructions of variable length each taking multiple cycles to carry out
8
Q
Uses of CISC and examples of manufacturers
A
- most desktops computer and laptops
- Intel (x86) and AMD
9
Q
CISC advantages
A
- simpler software (compiler does less work as emphasis is on hardware)
- larger instruction set = work better with more intensive tasks (executes complex instructions and carry out a number of low-level operations as part of a single instruction)
- more specialised instructions means low-level programs for CISC processors take up less space in memory and are easier to translate to machine code.
- small code sizes, higher cycles per second
-makes more efficient use of RAM (CISC requires less to store instructions)
-multiple addressing modes available
10
Q
CISC disadvantages
A
- More circuitry (transistors) needed to execute complex tasks as it aims to complete in as few lines of assembly as possible. (expensive)
-greater energy consumption, gives off more heat - instructions can take multiple machine/clock cycles to execute —> makes pipelining difficult therefore CISC processing used for specialist computer systems
11
Q
what is a co processor?
A
- any additional processor used for a specialised task
12
Q
- why did co processors emerge?
- what do they do?
A
- tech advanced —> demand increase on single CPU
- improve overall speed of computer by executing concurrently with main CPU, aided by a highly bespoke set of electronics
13
Q
outline the 2 key differences between CPU and GPU
A
- central processing unit is more general purpose in contrast to graphics processing unit which is highly specialised
- CPU excels at performing complex operations on small data sets but GPU excels at performing simple operations on larger data sets
14
Q
What is the GPU’s bespoke design useful for (PP). And what applications can have improved performance from use of this
A
- Parallel Processing
- improves performance of video/graphic-intensive applications
15
Q
Which contemporary architecture does it make use of and why. SIMD, MIMD or Distributed Computing?
A
- SIMD
- because mass calculation associated to graphics processing take full advantage of parallel processing
16
Q
what does SIMD stand for
A
- Simple Instruction Multiple Data
17
Q
What does the contemporary architecture SIMD allow for
A
- parallel processing when a processor carries out single instruction on multiple data items at the same time
18
Q
give some uses of GPU’S
A
- weather modelling
- data mining
- machine learning
- linear algebra
- statistical analysis
- cryptocurrency to mine bitcoins
- graphics rendering
19
Q
what is memory mapped output
A
- facility for producing some form of basic visual output by mapping values in memory to pixels on a screen
20
Q
what is the benefit of a separate, dedicated processor for a computer running applications that require video/graphic-intensive applications?
A
- this design frees up main processor to handle other tasks.
21
Q
what do graphics cards contain
A
- own set of RAM and processor (GPU)
22
Q
Although GPUs best known for use in gaming, what other specialised purposes can they be used for (hint: ML) and why?
A
- machine learning applications.
- GPUs with 1000’s of cores = optimised to parallel process calculations, required for machine learning algorithms (usually good for one task)
23
Q
give an example of machine learning
A
- training a neural network
24
Q
why are operations such as 3D graphics rendering divided across multiple cores?
A
- as each core performs a small part of the whole operation.
25
what are some main manufacturers of GPU’s
- Intel, NVidia, AMD
26
What’s involved in rendering 3D image
Example: calculations like RGB for individual pixels
- All segments of image assigned XYZ co-ordinates (requires 3D arithmetic, stored as floating point)
- Every segment assigned texture/pattern
- Light source assigned + given an angle, brightness + how it falls on wireframe
- Camera source chosen for image
27
what happens once calculations done?
- they’re run through every pixel of finished image from cameras perspective + generate 2D representation of 3D space
28
what is a multicore processor
- Single chip/processor containing 2 or more independent processing units which can independently process instructions simultaneously
29
what does CMPS stand for
- chip multiprocessors
30
What is CMPS?
- when manufacturers integrate these cores (multicore processor) onto single physical chip (CMPS)
31
why can be wrong to refer to single physical chip as CPU?
- as its actually a CMP containing multiple CPU cores
32
1. what is a CMP with 2 cores called
2. what is a CMP with 4 cores called
1. dual-core processor
2. quad-core processor
33
the performance of multicore processors is further enhanced by what 2 things?
1. on-chip shared cache
2. inter-core communication
34
explain why double cores doesn’t always double performance
- overheads involved with inter-core communication (they have to take their turn when accessing cache + RAM; all of which uses up some of the potential speed that adding more cores gives you)
- some programs cant make max use of all cores
35
Overall a multicore system should process _______ instructions _________ than a single core
1. multiple
2. faster
36
what is parallel processing
- processing of program instructions by dividing them between multiple processor cores
37
Improvement in _______________ gained by using multicore processors depends on ___________ taking advantage of ___________ ________________
1. performance
2. software
3. parallel
4. processing
38
the extent to which parallel processing can speed up computation will be highly dependent on what 2 things
1. task being carried out
2. if software has been/can be designed and written to make use of parallel processing
39
explain the benefit of parallel processing using multiple cores
- each processor (core) can execute a separate instruction at a time enabling multitasking as they process faster --> more processes completed per second
40
EXAM QUESTION:
Some problems require a large amount of computing power that goes well beyond single CPU. Discuss the different approaches that can be taken to provide increasingly larger amounts of computing power + the types of problem they are suited to.
AO1: Knowledge and Understanding
- Processors have increasingly large clock speeds and can be overclocked.
- Processors can have multiple cores.
- Super computers can have co and/or multiple processors
- GPUs can be applied to problems other than graphics processing.
- Problems can be distributed across a number of computers working together.
AO2.1: Application
- multiple cores can speed up smaller problems but won’t be enough for larger problems.
- Supercomputers are prohibitively exceptionally expensive to buy + run for all but large organisations.
- GPUs becoming cost efficient way of tackling problems as they tend to have large number of cores so can run on highly paralleliseable problems, only where same instruction being applied to multiple pieces of data (SIMD)
AO3.3: Evaluation
- Increased clock speed = limited to smaller problems. doubling clock speed would only halve time taken.
- Parallel processing isn’t suited to all problems as most are only partially parallelisable.
- GPUs suited to a subset of science/ engineering problems where same calculation repeated on multiple data sets.
• software (shader) that renders the display (1).
41
What is a parallel processor system?
- one which simultaneously uses several processors to perform single task which may be split into a number of sub-tasks each of which may be processed by any available processor.
42
what is sequential processing
- when instructions carried out one at a time in order which they are received
43
what’s a disadvantage of sequential processing
- very slow
