1.1.1c/d - Performance & Pipelining Flashcards
What is cache memory?
Temporary storage for frequently accessed data. Quick to access because it is closer to the CPU than RAM
What are the three levels of cache memory?
Level 1 is the smallest (less than a megabyte) but also the fastest
Level 2 is larger (up to 8 megabytes) but slightly slower
Level 3 is located outside of the CPU core which makes it slower than the first two, but is the largest (up to 50 megabytes)
What is a ‘cache hit’ and ‘cache miss’?
A ‘cache hit’ is when the CPU is searching for data and successfully finds it within cache.
A ‘cache miss’ is when data is not found in the cache and the CPU has to search in RAM (takes longer).
How does cache memory improve performance? (3 points)
- Cache memory is closer to the CPU than RAM - provides data and instructions quicker
- More cache memory should lead to higher performance as repeatedly used instructions can be stored and accessed faster
- Larger Level 1 and 2 cache will allow for data to be accessed extremely quickly.
What is the limitation of cache memory?
- Cache memory is costly so computers only have a small amount
- Multiple ‘cache misses’ will result in data latency as information is being accessed from the RAM
What is clock speed?
Clock speed is the measure of how quickly a CPU can process instructions.
How does clock speed improve performance
The faster the clock speed, the faster the computer can perform the FDE cycle resulting in better performance because more instructions can be processed every second.
What is overclocking?
Overclocking is when the computer’s clock speed is increased higher than the recommended rate. This will make the computer perform faster but could lead to overheating or could damage the machine.
What is underclocking?
Underclocking is when the computer’s clock speed is decreased lower than the recommended rate. This will make the computer perform slower but will increase the lifespan of the machine.
What is a core?
A core is a complete set of CPU components (CU, ALU and registers). Each core is able to complete it’s own FDE cycle.
How does the number of cores improve performance?
- In theory, a single-core processor can execute one instruction at a time, a dual-core processor can execute two instructions and so on.
- Therefore a computer with more cores will have higher performance because it can process more instructions at once.
What are the limitations of having more cores? ( 2 points)
- If one core is waiting for another to finish processing, performance might not increase at all.
- Some software is not written to make use of multiple cores, so it will not run any quicker on a multi-core computer.
What is pipelining?
Pipelining efficiently uses multiple cores or processors to perform multiple stages of the FDE cycle at the same time. This increases the amount of instructions that can be fetched, decoded and executed per second.
What is the limitation of pipelining?
Certain instructions may not be able to be executed until other instructions have been fetched and decoded. This wait for other instructions may impact performance.
