1.1.2 - Types of Processor

Question 1

RISC

Answer 1

Reduced Instruction Set Computer
- Small number of simple instructions
- Instructions completed in a single clock cycle
- So individual instructions are executed quickly, but many instructions are needed to complete a given task
- Lower power consumption and physically smaller, so best used in phones and the like
- Needs more registers to provide faster access to data
- Fewer addressing modes
- Processor hardware is simpler
- Only load and store instructions can access memory
- Tries to speed up execution by reducing the average clock cycles per instruction

Question 2

CISC

Answer 2

Complex Instruction Set Computer
- Supports a large number of complicated instructions
- Instructions may take many clock cycles to complete
- Operates at higher clock speeds than RISC systems
- Supports more addressing modes
- Tries to shorten execution time by reducing the number of instructions
- More complex processor design
- Many instruction types can access memory
- Individual instructions take longer to execute, but needs fewer instructions to complete a given task

Question 3

Multicore Systems

Answer 3

• Computer with multiple cores (processors)
• Work together, but perform different tasks
• Cores have their own CU, ALU, L1 cache and registers, but share L2 cache and RAM

Advantages:
- Increased speed and efficiency
- Allows for true multitasking
- More jobs can be done in the same time length because they are executed simultaneously
- Tasks can be shared to reduce the load on individual processors and avoid bottlenecks

Disadvantages:
- It is harder to programme for as you have to ensure each task has the correct input data
- Results from different processors need to be combined at the end, which adds stems
- Not all tasks can be split across processors
- Harder to debug because there are more variables
- Programmes have to be designed with the multi-core system in hand, otherwise it would not be able to make use of the extra functionality, and most programmes are not for maximum audience

Question 4

Parallel Systems

Answer 4

• Multiple processors work together to perform a single task
• The task is split up into smaller sub-tasks called Threads
• These Threads are split between processors and are then completed simultaneously, reducing the time it takes to complete the task
• Threads can be executed by any of the available processors
• Software has to be specially written to take place of these systems, which it often isn’t because companies want to make the most profit that they can by making products accessible to the lowest common spec
• All of the processors act as a standard core, loading instructions from memory and executing them as a regular processor would, so they could act independently
• However, they need to communicate together to keep edited data synchronised across the cores
• Controlled by the operating system

Question 5

Types of Pararallel Systems

Answer 5

MISD:
- Multiple Instruction, Single Data
- Each processor performs a different calculation on the same peice of data

SIMD:
- Single Instruction, Multiple Data
- Each processor performs the same calculation, just on different peices of data

MIMD:
- Multiple Instruction, Multiple Data
- Each processor performs a different instruction on a different peice of data
- Allows for true multitasking

Question 6

Co-Processors

Answer 6

• The additional processor could not operate independently as it is specialised to one specific, repetitive task only
• The additional processor is responsible for performing a specific task, and that task only
  — Ie. graphics or floating point arithmetic
• The main processor and co-processor operates on different tasks at the same time, speeding up tasks by taking some of the burden from the main CPU

Question 7

GPUs

Answer 7

• Graphics Processing Units
• Dedicated co-processor for performing the calculations to render graphics
• Each GPU has many parallel cores to allow for multiple areas to be processed at once, increasing speed
• Require their own heat sink because of the amount of calculations that they are doing
• SIMD processing, performing the same, repetitive rendering calculations on all objects in a scene

Question 8

Overclocking

Answer 8

• Increasing the clock speed to higher than it is supposed to go to
• Allows calculations to be performed faster, and therefore improves performance
• However, this produces more heat, which is damaging and reduces the lifespan of the components
• And takes more power, so is more expensive to run