CPU components and performance
A modern CPU is built from a small number of named subsystems. AQA expects you to know each one and to discuss how clock speed, cache and number of cores affect performance.
CPU components
ALU (Arithmetic Logic Unit)
Performs arithmetic (add, subtract, multiply, divide) and logical operations (AND, OR, NOT, comparisons). It receives two operands from registers, performs the operation and writes the result back to a register.
Control Unit (CU)
Decodes instructions and orchestrates the rest of the CPU: tells the ALU which operation to perform, signals memory to read/write, manages the flow of instructions. The conductor of the CPU orchestra.
Registers — fast on-chip storage
A handful of tiny storage locations the CPU can access in a single cycle. Key registers:
- Program Counter (PC) — address of the next instruction.
- Memory Address Register (MAR) — address being read or written.
- Memory Data Register (MDR) — data being transferred to/from memory.
- Current Instruction Register (CIR) — current instruction being decoded.
- Accumulator ACC — holds intermediate arithmetic results in simple architectures.
Registers are far faster than RAM but have very small capacity (a few words).
CPU performance factors
Clock speed
Measured in hertz (Hz) — one Hz = one cycle per second. Modern CPUs run at 2-5 GHz (billions of cycles per second).
- Higher clock speed → more cycles per second → more instructions per second (other things equal).
- Trade-off: more heat, more power consumption, harder to cool.
- "Overclocking" pushes a CPU beyond its rated speed; can cause instability.
Cache memory
A small, very fast memory inside or near the CPU that stores recently used data and instructions. When the CPU needs the same data again, the cache satisfies it without going to main RAM (much slower).
Hierarchy:
- L1 cache — smallest, fastest, on-CPU. Typically 32-64 KB.
- L2 cache — larger, slightly slower. ~256 KB - 1 MB.
- L3 cache — shared across cores, larger again. ~4-30 MB.
Larger / more cache → fewer slow trips to RAM → better performance, especially on data-heavy workloads.
Number of cores
A core is a complete CPU on a single chip. Multi-core processors:
- 2 cores → can run two processes simultaneously.
- 4-8 cores → typical desktop/laptop.
- 16+ cores → workstations and servers.
More cores helps when:
- Multiple programs run at once.
- A program is multi-threaded (split into parallel tasks).
More cores helps less when:
- A single-threaded program can't use multiple cores.
Comparing two CPUs (exam pattern)
If a question gives two CPUs:
- Higher clock speed → faster on single-threaded work.
- More cores → better at multitasking and parallel workloads.
- More cache → fewer RAM trips, better with large data.
- Higher RAM capacity (separate from CPU) → larger programs / datasets.
You can rarely declare one CPU "better overall" — it depends on the workload.
Real example numbers
| CPU | Clock (GHz) | Cores | L3 cache |
|---|---|---|---|
| A | 3.0 | 4 | 8 MB |
| B | 2.5 | 8 | 16 MB |
A is better for a single-threaded game; B is better for video editing or running many apps.
✦Worked example
A pupil's PC has 2 cores at 2.0 GHz and 2 MB cache. Suggest two upgrades and why.
- Upgrade to more cores → faster multitasking, better at parallel workloads B1.
- Upgrade to higher clock speed → each core runs faster B1.
- More cache → reduces time spent waiting on RAM B1.
⚠Common mistakes— Pitfalls
- More clock speed = always faster — not if other factors (cache, branch prediction) suffer.
- Confusing cores with threads. Hyper-threading lets one core run two threads, but they share resources.
- Treating cache as RAM. Cache is much smaller and faster — mediates between CPU and RAM.
- Forgetting that not all programs use all cores. A single-threaded game on a 16-core CPU uses one core.
- Saying "more cores always better". Diminishing returns, especially for sequential workloads.
➜Try this— Quick check
For each task, which CPU upgrade helps most?
- A web browser opening many tabs → more cores / more RAM.
- A single-threaded chess engine → higher clock speed.
- Video encoding → more cores + bigger cache.
- General responsiveness → bigger / faster cache and reasonable clock.
AI-generated · claude-opus-4-7 · v3-deep-computer-science