Processor Performance
There are 3 main factors that affect CPU performance:
1. Clock speed
2. The number of cores, or duplicate cores, linked together on a single chip
3. The amount and type of cache memory
Clock speed:
- Signals are generated by the system clock which synchronises CPU operations
- Each CPU operation starts as the clock changes from 0 to 1 or 1 to 0
- The clock cycle caps the amount of operations performed per unit time
- One clock cycle is 1 Hz with clock speed being measured in GHz
- All processor activities begin on a clock pulse
Number of cores:
- Dual-core and quad-core processors has multicores linked together
- In theory, each processor can individually process a separate instruction at a time
- So, the processing power is doubled or quadrupled
- But the software must be designed to allow this otherwise it won’t use this ability to
its advantage
Amount and type of cache memory:
- Cache is very small, fast memory located inside the CPU chip
- It provides quicker access time to instruction as it is easier to access than the main
memory, so the average speed is increased
- As cache fills up, unused instruction or data still being held are replaced with more
recently used ones
- There are 3 levels:
1. Level 1 – fastest of the 3 but smallest, 2-63KB
2. Level 2 – fairly fast and fairly large, 256KB – 2MB
3. Level 3
Pipelining:
- Pipelining improves CPU performance by processing instructions simultaneously
- Usually, the fetch-decode-execute cycle happens sequentially. I.e. while the next
instruction is being fetched, the ALU is idle
- So, pipelining allows the processor to fetch the next instruction while the current
one is being operated on, holding them in a buffer close to the CPU until the
instruction can be performed
- Pipelining is split into two ‘pipes’, the instruction pipeline and the arithmetic
pipeline
- The instruction pipeline is split into the stages an instruction would go through:
fetched, then buffered and then executed
- The arithmetic pipelining is split into the parts of an arithmetic operation that can be
broken down and overlapped as they are performed
There are 3 main factors that affect CPU performance:
1. Clock speed
2. The number of cores, or duplicate cores, linked together on a single chip
3. The amount and type of cache memory
Clock speed:
- Signals are generated by the system clock which synchronises CPU operations
- Each CPU operation starts as the clock changes from 0 to 1 or 1 to 0
- The clock cycle caps the amount of operations performed per unit time
- One clock cycle is 1 Hz with clock speed being measured in GHz
- All processor activities begin on a clock pulse
Number of cores:
- Dual-core and quad-core processors has multicores linked together
- In theory, each processor can individually process a separate instruction at a time
- So, the processing power is doubled or quadrupled
- But the software must be designed to allow this otherwise it won’t use this ability to
its advantage
Amount and type of cache memory:
- Cache is very small, fast memory located inside the CPU chip
- It provides quicker access time to instruction as it is easier to access than the main
memory, so the average speed is increased
- As cache fills up, unused instruction or data still being held are replaced with more
recently used ones
- There are 3 levels:
1. Level 1 – fastest of the 3 but smallest, 2-63KB
2. Level 2 – fairly fast and fairly large, 256KB – 2MB
3. Level 3
Pipelining:
- Pipelining improves CPU performance by processing instructions simultaneously
- Usually, the fetch-decode-execute cycle happens sequentially. I.e. while the next
instruction is being fetched, the ALU is idle
- So, pipelining allows the processor to fetch the next instruction while the current
one is being operated on, holding them in a buffer close to the CPU until the
instruction can be performed
- Pipelining is split into two ‘pipes’, the instruction pipeline and the arithmetic
pipeline
- The instruction pipeline is split into the stages an instruction would go through:
fetched, then buffered and then executed
- The arithmetic pipelining is split into the parts of an arithmetic operation that can be
broken down and overlapped as they are performed
