Skip to topic
    ← Back to course topics

    Systems architecture — AQA GCSE Computer Science

    Verified courseTeaching note checked against the official specification.

    Systems architecture defines the structure and behaviour of computer systems, focusing on the Von Neumann architecture, CPU components, and the Fetch-Decode-Execute cycle. Understanding how clock speed, cache size, and processor cores influence performance is vital for mastering AQA GCSE Computer Science Paper 1.

    Systems architecture explained

    The fetch stage is the first part of the fetch–decode–execute cycle.

    Read the full explanation

    The CPU needs the next instruction before it can do anything with it, and instructions are held in main memory (RAM) alongside data. The program counter (PC) holds the address of the next instruction; the CPU copies that address along the address bus to memory, and memory returns the instruction along the data bus into the current instruction register (CIR). The PC is then incremented so it points to the following instruction. So 'fetch' means moving the next instruction from main memory into the CPU, not carrying it out and not moving data to be stored. A useful check: if the question describes an address being sent and an instruction coming back, that is fetch.

    Read the Systems architecture study guideFull revision notes for AQA GCSE Computer Science

    Your focus

    1. Identify fetch as the stage that brings the next instruction from main memory into the CPU.
    2. Describe the role of the program counter in supplying the instruction address.
    3. Distinguish fetch from decode and execute.

    Which command word is this?

    Provide a brief, factual answer without explanation or expansion (e.g., 'State the register that holds the result of calculations: Accumulator').

    Systems architecture exam tips

    Topic Overview

    Systems architecture explores the internal design of the central processing unit (CPU) and how hardware components cooperate to process data. It centres on the Von Neumann model, exploring registers, buses, and the fundamental Fetch-Decode-Execute cycle that underpins every digital computer.

    Understanding systems architecture enables students to explain how hardware specifications, including clock speed, cache memory, and multi-core processing, directly affect software performance. This foundational topic links directly to memory, secondary storage, and low-level software operations in AQA GCSE Computer Science Paper 1.

    Key Concepts
    • →Von Neumann Architecture: Uses shared memory for both program instructions and program data, linked to the CPU via address, data, and control buses.
    • →CPU Registers: Dedicated high-speed storage locations inside the CPU, including the MAR, MDR, PC, and Accumulator (ACC).
    • →The Fetch-Decode-Execute Cycle: The continuous sequential cycle where instructions are retrieved from RAM, interpreted by the CU, and executed by the ALU or registers.
    • →CPU Performance Factors: System performance is determined by the interaction between clock speed (Hz), cache capacity/levels (L1, L2, L3), and the number of processor cores.
    • →Embedded Systems: Dedicated computers built into larger physical machines designed to perform specific, repetitive control tasks with limited hardware resources.
    Marking Points
    • Fetch is the first stage of the fetch–decode–execute cycle.
    • The next instruction is copied from main memory (RAM) into the CPU.
    • The program counter holds the address of the next instruction, which is sent to memory.
    • The fetched instruction is placed in the current instruction register (CIR).
    • The program counter is incremented to point to the following instruction.
    • Fetch does not execute or decode the instruction; those are later stages.
    Examiner Tips
    • 💡Link each stage to a register: PC supplies the address, CIR holds the fetched instruction.
    • 💡Use the phrase 'next instruction' to show you know fetch looks ahead, not at the current one.
    • 💡If asked to order stages, put fetch first, then decode, then execute.
    • 💡When naming registers, always use the full formal term alongside the acronym at least once (e.g. 'Memory Data Register (MDR)') to guarantee credit.
    • 💡In questions asking about clock speed, explicitly include the unit 'Hertz' and define it as 'cycles per second' rather than 'operations per second'.
    • 💡For multi-core comparison questions, explicitly use the term 'parallel processing' and mention the constraint of 'instruction dependencies' to reach top-band marks.
    Common Mistakes
    • Confusing fetch with execute: the correction is that fetch only brings the instruction into the CPU; carrying it out is the execute stage.
    • Saying the instruction is fetched from the hard disk or cache: the correction is that the next instruction is fetched from main memory.
    • Thinking fetch stores a result back to memory: the correction is that fetch reads the instruction in; writing results back is a later step.
    • Believing that Cache memory stores long-term files like a hard drive; cache is volatile, extremely fast SRAM located on or next to the CPU to store frequently used instructions and data.
    • Thinking that 'clock speed' means how many instructions are completed per second; clock speed measures clock cycles per second, and some instructions take multiple cycles to complete.
    • Assuming an embedded system is simply any small computing device like a smartphone; smartphones are general-purpose computers, whereas embedded systems (e.g., washing machine controllers) execute dedicated firmware.
    Revision Plan
    1. 1Step 1: Memorise the Von Neumann diagram, labelling the CU, ALU, registers (PC, MAR, MDR, ACC), and system buses accurately.
    2. 2Step 2: Write out the step-by-step Fetch-Decode-Execute cycle using bullet points, ensuring register transfers are clearly stated.
    3. 3Step 3: Create comparison flashcards detailing how clock speed, cache size, and core count specifically improve or constrain CPU performance.
    4. 4Step 4: Practice distinguishing between general-purpose systems and embedded systems with five concrete real-world examples.
    5. 5Step 5: Complete timed past-paper questions focusing on 4-mark and 6-mark explain/evaluate stems from recent AQA Paper 1 papers.
    Exam Question Types
    • 📋Register identification questions: Short 1-2 mark questions asking which register holds an address, data, or calculation output during a specific stage.
    • 📋Cycle trace explanations: 3-4 mark questions requiring an ordered description of the Fetch, Decode, or Execute stages.
    • 📋Performance evaluation scenarios: 4-6 mark extended responses evaluating the effect of upgrading specific CPU components for given user requirements.
    • 📋Embedded systems characteristics: 2-3 mark questions requiring students to define an embedded system and identify its operational benefits (e.g., low power, reliability).
    Command Word Expectations (AQA)
    State / Give

    Provide a brief, factual answer without explanation or expansion (e.g., 'State the register that holds the result of calculations: Accumulator').

    Explain

    Give reasons or causes for a concept. Use connective phrases like 'because', 'which leads to', or 'therefore' to connect hardware attributes to their direct functional impact.

    Evaluate

    Consider multiple factors, weigh up advantages and limitations of hardware choices in relation to a specific scenario, and provide a justified, balanced conclusion.

    How Students Lose Marks (Examiner Pitfalls)
    Pitfall: Confusing the roles of the Program Counter (PC) and Memory Address Register (MAR) during the fetch stage.
    ❌ Weak Answer (Loses Marks):The PC gets the instruction from RAM and gives it to the MAR so it can run.
    Example improved answer:The Program Counter (PC) holds the address of the next instruction to be fetched. This memory address is copied across the address bus to the Memory Address Register (MAR). The PC is then incremented by one to point to the next sequential address.
    Examiner Tip: Explicitly state that the PC holds an 'address', not the instruction itself, and mention that the address is copied to the MAR before the PC increments.
    Pitfall: Stating that doubling CPU cores will automatically double system performance under all circumstances.
    ❌ Weak Answer (Loses Marks):If a CPU has four cores instead of two, it works twice as fast because it can do twice as much work at the same time.
    Example improved answer:Doubling cores increases potential parallel processing capacity, but performance does not double linearly. Some software cannot be parallelised into concurrent threads, and performance is constrained by inter-core communication overhead and sequential task dependencies.
    Examiner Tip: Always mention dependencies between instructions and software multithreading capability when discussing multi-core performance.
    Step-by-Step Worked Solutions

    Question: Explain the role of the Arithmetic Logic Unit (ALU) and the Control Unit (CU) during the Fetch-Decode-Execute cycle. [4 marks]

    1. 1.Step 1: Define the primary function of the Control Unit (CU) in terms of overall CPU coordination.
    2. 2.Step 2: Describe the CU's specific duty in the cycle, such as decoding instructions and sending control signals to components.
    3. 3.Step 3: Define the primary function of the Arithmetic Logic Unit (ALU) regarding data manipulation.
    4. 4.Step 4: Detail the operations performed by the ALU, specifically mathematical calculations (e.g. addition, subtraction) and logical comparisons (e.g. AND, OR, NOT).
    Final Answer: The Control Unit (CU) coordinates CPU operations, manages the Fetch-Decode-Execute cycle, decodes instructions stored in the Current Instruction Register (CIR), and directs data flow using control signals. The Arithmetic Logic Unit (ALU) carries out arithmetic operations (such as addition and subtraction) and performs logical comparisons and Boolean operations (such as AND, OR, and comparisons of magnitude).

    Question: A dual-core processor runs at 3.0 GHz, while a quad-core processor runs at 2.4 GHz. Evaluate which CPU would be more suitable for running video editing software that supports parallel processing versus legacy administrative software that runs on a single thread. [6 marks]

    1. 1.Step 1: Analyse the video editing scenario in relation to core counts and multithreading.
    2. 2.Step 2: Calculate or compare the aggregate theoretical throughput for parallel tasks (e.g., 4 x 2.4 GHz vs 2 x 3.0 GHz).
    3. 3.Step 3: Analyse the legacy software scenario, which can only execute on a single core.
    4. 4.Step 4: Compare single-core clock speeds (3.0 GHz vs 2.4 GHz) for single-threaded performance.
    5. 5.Step 5: Provide a clear, justified conclusion for both software types.
    Final Answer: For video editing software engineered for multithreading, the quad-core 2.4 GHz processor is more suitable. Parallel tasks can be distributed across four cores, yielding higher overall throughput despite the slightly lower clock speed. Conversely, the legacy administrative software cannot take advantage of multiple cores and executes on a single core. For this application, the dual-core 3.0 GHz processor will perform better, because its individual core executes 3.0 billion cycles per second compared to 2.4 billion cycles per second on the quad-core chip.
    Active Recall Memory Test
    What specific value is held inside the Accumulator (ACC)?
    Key Fact: The Accumulator holds the temporary result of the most recent arithmetic or logic operation performed by the ALU.
    Which bus carries the physical memory address from the CPU to the RAM?
    Key Fact: The address bus.
    Why does increasing L1/L2 cache size improve CPU performance?
    Key Fact: Cache provides faster access speeds than RAM, reducing idle CPU wait cycles by keeping frequently used data immediately accessible.
    Give two typical characteristics of an embedded system.
    Key Fact: Dedicated to a single specific function, has limited memory/processing power, consumes low power, and runs firmware without an interactive OS.
    Frequently Asked Questions
    What is the difference between the MAR and the MDR?
    The Memory Address Register (MAR) stores the memory location (address) in RAM where data or an instruction needs to be read from or written to. The Memory Data Register (MDR) holds the actual data or instruction that has just been fetched from RAM or is waiting to be written to RAM. In short: MAR holds 'where' in memory, while MDR holds 'what' was retrieved or is being stored.
    Why doesn't a 4.0 GHz CPU run four times faster than a 1.0 GHz CPU?
    Clock speed measures how many machine cycles occur per second, but overall processing speed depends on multiple bottlenecks. The CPU frequently waits for data to arrive from slower RAM (known as memory latency). Furthermore, complex instructions can take multiple clock cycles to finish, and thermal throttling may force the CPU to decrease its speed to prevent overheating.
    Is an iPad or a smartphone considered an embedded system?
    No, an iPad or smartphone is not an embedded system; it is a general-purpose computer. It has an operating system capable of downloading, installing, and executing many different applications for diverse tasks. An embedded system, like the microcontroller in an anti-lock braking system (ABS) or microwave, is dedicated exclusively to one single predetermined control task.
    What is the Von Neumann architecture bottleneck?
    The Von Neumann bottleneck occurs because program instructions and data share the same physical bus system and memory space. Because data throughput between the CPU and memory is significantly slower than the rate at which the CPU can execute instructions, the CPU spends idle clock cycles waiting for data or instructions to arrive over the bus.
    How do L1, L2, and L3 cache differ from each other?
    L1 cache is the smallest, fastest, and most expensive memory, built directly inside each individual CPU core. L2 cache is slightly larger and marginally slower than L1, often dedicated to each core. L3 cache is larger still and slower than L1 and L2, but it is typically shared across all processor cores on the CPU chip to avoid fetching repeated data from system RAM.