3.3.3: Computer architectures

3.3.3: Computer architectures

1. Von Neumann

   • Takes each instruction serially
   • Performs each instruction before carrying out the next.
   • Instructions and data stored together in memory

2. FETCH

   • 1: PC holds address of next instruction to be executed
   • 2: Address is copied to the MAR
   • 3: Data in address at MAR is copied to MDR
   • 4: Contents of MDR copied to CIR
   • 5: Contents of PC are incremented

3. DECODE

   • 6: Contents of CIR divided into code for operation to be carried out and the address that the data will be used by
   • 7: Control Unit interprets the code so that the processor knows what to do next.

4. EXECUTE

   • 8: Address is copied from the CIR to the MAR
   • 9: Data found in the MAR is copied to the MDR
   • 10: Data is used

5. Types of processor

   • Co-processor
     • Additonal processor for a specific task (such as a GPU)
     • Most co-processors do not fetch instructions for themselves
   • Parallel Processor
     • With Fetch - Decode - Execute cycle
       • A processor for each step of the Fetch - Decode - Execute cycle
       • Speeds up process as next instruction can be fetched before current instruction has finished processing.
     • With Pipelining
       • One main processor that splits jobs up into tasks and farms each tasks to other processors
       • Main Processor Fetches and Decodes task and splits execution into individual tasks and sends to other processors to execute.
   • Array processor
     • Multiple processors compute mathematics on arrays of numbers simultaneously.
     • Numbers are all fetched and then executed simultaneously.
   • Advantages and Disadvantages of Alternative Processors
     • Advantages
       • Faster Processing
       • More than one instruction can be executed at the same time
       • Different processors can handle different parts of the same job.
     • Disadvantages
       • Requires processors to be synced together
       • Program has to be written to use multiple processors
       • Programs are more complex to write, test and debug.

6. CISC vs RISC

   • CISC
     • Every computer operation has an instruction
     • Each operation has a binary code (Op-Code) associated with it
     • Complex Instruction Set
     • More bits needed to represent each instruction
   • RISC
     • Reduced Instruction Set
       • Typically used in handheld devices
     • Recognises fewer instructions
     • Less bits needed to represent instructions
     • Missing instructions made up with programming techniques
     • Less complex circuitry
     • Op-code and address allocated same number of bits
     • Registers are all the same size

Images and files in the map