Processor Performance

• Words

  Memory is divided into equal units called words. Each word has its own memory address. Word length is usually 8, 16, 32 or 64 bits.
• Address Bus

  The width of the address bus determines the maximum possible memory addresses of the system. An average PC has a memory capacity of 4GiB which is 2^32 bytes aka a 32 bit address bus.
• Clock Speed


  The fetch-execute cycle is triggered by the clock pulses of the system clock. The faster the clock speed, the faster a computer can fetch, decode and execute instructions. The clock can change state many billions of times per second. A 2GHz processor would tick 2 billion times per second.
• Data Bus


  It is bi-directional as data can be sent both ways along the bus. If the data is the same width as a computer word, data can be transferred to and from memory in a single operation. The width of the data bus is defined by the number of wires or lines it contains.
• Factors Affecting Performance

  • clock speed
  • number of cores
  • amount and type of cache memory
• The System Clock

  A series of ON/OFF signals are used to synchronise the operations of the processor components. Action are usually carried out on the rising edge of the clock. Actions each take a fixed number of cycles to complete.
• Cache Memory

  Cache is a small amount of superfast, expensive memory that stores data and instructions that have recently bee used by the processor. The more cache memory a computer has the more likely it will be that it will not have to fetch the next instruction or data from RAM, as it will already have been loaded into the superfast cache memory from which it can be retrieved much more quickly.
• Level 1 Cache


  Level 1 cache is split into data cache and instruction cache so that data and instructions can be fetched simultaneously.
• Parallel Processing


  It is when several processor cores are working at the same time. Each core can work concurrently on different parts of the same task. As instructions are processed sequentially this is not always possible.
• Number of Cores

  A dual core computer has 2 processors linked together in the same integrated circuit. Each core is theoretically able to process a different instruction at the same time with its own fetch-execute cycle, making a quad core 4x faster than a single core processor. However, the software may not always be able to make full advantage of all 4 processors.
• Pipelining


  Improves performance by overlapping stages of the fetch-execute cycle or by breaking down stages in an arithmetic operation. There may be 10 or 12 stages in the pipeline with some taking longer than others.