Computer systems and networks · GCSE Computer Science
Systems architecture
The CPU’s fetch–decode–execute cycle, what actually speeds a processor up, memory versus storage, and embedded systems.
Fetch, decode, execute — every instruction, every cycle.
The important bits
What you need to know
- 1
Von Neumann machines store programs and data in the same memory. The CPU fetches instructions from RAM, decodes them, executes them, then repeats. That loop is the fetch–decode–execute cycle.
- 2
ALU performs arithmetic and logic. The control unit orchestrates the cycle and control signals. Registers are tiny, fast storage inside the CPU (including the program counter, accumulator/accumulator-like registers, instruction register — names vary by board).
- 3
Cache is small, fast memory close to the cores holding recently used instructions/data. More cache often means fewer slow trips to RAM. It is not the same as RAM, and it is not storage.
- 4
Clock speed is cycles per second (GHz). A faster clock can mean more instructions per second, until the CPU waits for memory or heat wins. Cores allow true parallel instruction streams for software that can use them. Three factors together: clock, cores, cache.
- 5
RAM is volatile primary memory: currently running programs and data. ROM is non-volatile firmware (bootstrap). Virtual memory uses disk as an overflow for RAM — it prevents crashes from shortage and can thrash if overused.
- 6
Secondary storage keeps data when power dies: magnetic (HDD: cheap, moving parts), solid state (SSD/flash: fast, no moving parts, limited write cycles), optical (discs: cheap distribution, slow, scratchable). Compare with capacity, speed, portability, durability, reliability, cost.
- 7
Embedded systems are computers inside a function: dishwashers, cars, medical devices. They have dedicated software, often real-time constraints, and limited UI. Your phone is a general-purpose computer that also contains embedded controllers.
- 8
Cloud storage is someone else’s disk accessed over a network. Benefits: access from anywhere, someone else handles hardware. Costs: bandwidth, ongoing fees, jurisdiction, and the need to trust the provider’s security.
Go deeper
What the clock does not magically fix
Students write “higher clock speed = faster computer” and stop. Each tick is a chance to move the cycle along, but an instruction that waits for RAM, a disk, or a locked core is not helped by a louder metronome. That is why cache exists: keep the working set near the ALU. Extra cores help a video encode or a busy server; they do little for a single-threaded homework script. Heat and power cap clocks in laptops. A 4-mark “explain how cache / cores / clock affect performance” wants three separate mechanisms, not the word “faster” three times. If the question gives a use case (gaming, embedded sensor), pick the bottleneck that actually matters.
Go deeper
Memory is not storage
RAM forgets at power-off; that is why you save. ROM does not forget, but you do not keep your essay there. Virtual memory copies pages to disk so more programs can appear loaded than RAM allows; if the disk thrashes, the “computer is slow” complaint is a memory-size problem in disguise. Secondary storage is the hierarchy below: HDD versus SSD is a trade of pence-per-gigabyte against latency and shock resistance. Optical media still appears in exams as distribution, not as a serious laptop drive. When comparing, use the six-word grid: capacity, speed, portability, durability, reliability, cost. A sentence for two of those beats a brand name.
See the idea in action
Compare HDD and SSD for a school laptop. SSD: faster boot and app load, more durable in bags, quieter, more expensive per GB, finite writes (rarely the student’s problem). HDD: cheaper bulk, slower, vulnerable to drops. Recommendation: SSD for the OS and working files; perhaps cloud or a cheap HDD for archive. That evaluated comparison is a 6-mark shape: criteria, not fanboy.
Exam technique
Turn knowledge into marks
Learn one clean FDE description: PC holds next address → fetch instruction into IR → decode in CU → execute in ALU or memory → PC updates. If you mix RAM into the ALU you lose the picture.
Common mistakes
Do not give these marks away
- 01
Calling RAM or cache “storage”, or saying files “live in the CPU”.
- 02
Thinking more cores always double speed for every program.
- 03
Writing that virtual memory is the same as cloud storage.
What is the role of the program counter?
ATo store the entire operating system
BTo hold the address of the next instruction to fetch
CTo draw graphics on the monitor
DTo encrypt files on the hard drive
Show the answer
To hold the address of the next instruction to fetch. Each cycle, the CPU needs to know where the next instruction lives in memory. The program counter stores that address and is updated as instructions run (including jumps).
Quick questions
If this is the bit you searched
Is cache just a smaller RAM?
It is closer, faster, and organised to hold copies of what the CPU is using now. You cannot store your photos there as if it were a tiny SSD. When cache misses, the CPU waits on RAM.
Why do embedded systems often use flash or ROM?
They must boot reliably, often without a disk, and the software rarely changes. A dishwasher does not need a 2 TB HDD; it needs cheap, robust firmware.