Flashcards
Types of processor
Cards in this deck (27)
- 1Stored program conceptMachine code instructions are fetched and executed serially by a processor that performs arithmetic and logical operations
- 2Conditions for stored program concept1. A program must be resident in main memory to be executed
2. The machine code instructions are fetched from memory sequentially, decoded and executed in the processor - 3Von Neumann architectureInstructions and data are stored in the same memory unit and share the same address and data buses
All memory has the same word length, regardless of its type (data or instructions) - 4Advantages of Von Neumann architecturePrograms can be optimised in size
Less wiring and therefore cheaper - 5Disadvantages of Von Neumann architecturevon Neumann bottleneck - bandwidth between CPU and memory is very small in comparison with the volume of data to be transferred, which limits CPU performance
- 6Harvard architecturePhysically separate memories and address/data buses for instructions and data
Memories may be different types - e.g. read-only for instructions and read-write for data
Memories may have different word sizes - some instruction sets are large and so have - 7Advantages of HarvardFaster CPU as instructions and data are fetched in parallel
- 8Disadvantages of HarvardMore wiring so more expensive
Programs tend to be larger - 9Harvard/Von Neumann usage comparisonHarvard: Tends to be used in embedded/DSP/audio systems
v
Von Neumann: Tends to be used in conventional PCs, servers and embedded systems with only control functions - 10Contemporary processor architecturesModern chips incorporate aspects of von Neumann and Harvard architectures
Examples:
1. Main memory holds data and instructions, but the cache is divided into memory and data cache
2.Some DSP processors e.g. Ti TMS320 C55x have multiple parallel data buses and one instruction bus - 11CISCComplex Instruction Set Computer
Has a very large number of instructions in its instruction set
Much of the decomposition of the task occurs in the processor's hardware and so each instruction may take multiple clock cycles to complete - 12Advantages of CISCPrograms are smaller - more can be added to RAM at once
Less work for the compiler as instruction set is similar to high-level code - 13Disadvantages of CISCMany specialised instructions - only 20% of instructions are used in an average program
More transistors are required to decode individual instructions, increasing production cost and power consumption
Instructions take multiple clock cycles so cannot be pipelined - 14Uses of CISCGeneral use PCs (x86 and AMD64 are both CISC)
- 15RISCReduced Instruction Set Computer
Has a very small number of instructions in its instruction set
Each instruction takes roughly one clock cycle to complete - 16Advantages of RISCHardware does very little work to decode instructions, resulting in cheaper production and lower power consumption
Instructions can be pipelined, making processing faster and more efficient - 17Disadvantages of RISCMore work for the compiler to translate high-level code
Programs tend to be larger so fewer can be loaded into RAM at once - 18Uses of RISCMobile devices (e.g. smartphones, tablets)
Some embedded systems (ARM and DSP processors) - 19Co-processorsA co-processor is an additional processor alongside the CPU which serves to supplement its functions
They tend to be quite specialised - 20Examples of co-processorsGraphics Processing Unit (GPU)
Sound
Digital signal processors
Sound processing
Floating-point arithmetic - 21Scalar CPUPerforms computations on a single number or dataset before moving on to the next
May be known as Single Instruction Single Data (SISD) - 22Array/Vector CPUAlso known as Single Instruction Multiple Data (SIMD)
Applies a single instruction to multiple data locations simultaneously - which is often much faster - 23Parallel ProcessorUses independent processors to work on the same program
Process is split into tasks which may be processed by any of the processors - 24Advantages of parallel processingFaster than a single processor for lots of data
Not limited by bus transfer rate - 25Disadvantages of parallel processingNot every problem can be decomposed to take advantage of parallel architecture
More expensive - additional processing blocks are required
Software written for it must be written to take advantage of the architecture, making it harder to write
The OS must be complex enough to communicate with all the processors - 26GPUsGraphics Processing Unit
Co-processor designed for graphics rendering
Operate in SIMD with thousands of cores, allowing the same instruction to be performed on many data points - 27Uses of GPUsGraphics rendering
Password cracking
Machine learning
Audio processing
Image processing
Financial transactions
Oil exploration
System modelling
Information
- Description
- A level OCR A level Computer Science (Topic 1 - Components of a Computer) Flashcards on Types of processor, created by James Norman on 21/01/2020.
- Created by
- James Norman
- Views
- 1
- Rating