A-level Computer Science revision · Processors, software and development

CPU performance and pipelining

The idea

CPU performance depends on clock speed (cycles per second), number of cores and cache size. Pipelining overlaps the fetch, decode and execute stages of different instructions, so once full, an ideal pipeline finishes one instruction per cycle. Branches can flush the pipeline. Von Neumann architecture shares memory for data and instructions; Harvard keeps them separate.

  1. Higher clock speed: more instructions per second.
  2. More cores: more tasks in parallel (if software allows).
  3. Pipelining: overlap stages; branches cause flushes.

Watch out Doubling cores does not double speed if the software cannot run in parallel.

See it worked

Worked example

Why does a branch instruction reduce the benefit of pipelining?

  1. The next instructions were fetched assuming no branch.
  2. If the branch is taken, they are wrong.
  3. The pipeline is flushed and refilled.

Answer The pipeline must be flushed and refilled

Your turn

Question 11 mark

Which architecture stores data and instructions in separate memories?

Choose one
Question 21 mark

Why does a larger cache improve performance?

Choose one
Question 32 marks

A non-pipelined processor takes 3 cycles per instruction. How many cycles does it need for 10 instructions?

A-level Computer Science Pro, OCR H446

A-level Computer Science Pro

£119 once

Price rises to £199 on 1 January. Save £80 by joining before then.

At least 12 months of access

By paying you agree to our Terms and get access straight away, which ends your 14-day cancellation right.

No VAT is charged.

Independent practice for OCR A-level Computer Science (H446), not endorsed by OCR.

Privacy · Terms