Systems architecture: number of processor cores

The idea

The number of processor cores is how many independent processing units sit inside a CPU, each able to fetch, decode and execute its own stream of instructions. A single-core processor works through one instruction stream at a time; a multi-core processor can run several streams in parallel, so tasks that can be split up (such as rendering video frames or serving many web requests) may finish sooner. More cores do not automatically make every program faster: software must be written to divide work between cores, and cores often share cache and memory bandwidth, so gains are not unlimited. Clock speed, cache size and core count together shape performance.

What earns the marks

• States that a core is an independent processing unit within the CPU that can fetch, decode and execute instructions.
• Explains that multiple cores allow several instruction streams or tasks to be processed in parallel.
• Recognises that more cores can improve performance for suitable multitasking or parallelisable software.
• Notes the limitation that performance gain depends on software using the cores and on shared resources such as cache and memory bandwidth.
• Distinguishes core count from clock speed and cache size as separate factors affecting CPU performance.

Common mistakes and how to put them right

• Confusing the number of cores with clock speed.

  Put it right: Cores are separate processing units, whereas clock speed is cycles per second of a processor.

• Assuming more cores always makes any program faster.

  Put it right: The software must be able to split work across cores to benefit.

• Treating cache size and core count as the same thing.

  Put it right: Cache is fast memory close to the CPU; cores are processing units.

Exam tips

• Read the question stem to see whether it asks what a core is or what effect more cores have, then choose accordingly.
• Eliminate options that describe clock speed or cache when the question is about cores.
• Remember the limitation about software and shared resources, as distractors often ignore it.