Learning Objectives
- Explain how clock speed, number of cores and cache size can affect CPU performance.
- Recognise that increasing one processor specification does not guarantee the same improvement in every task.
- Explain the purpose of an instruction set.
- Describe the relationship between machine-code instructions and processor architecture.
- Evaluate processor choices using more than one relevant factor.
Key Terms
- Clock speed
- The number of processor clock cycles completed per second, normally measured in hertz.
- Hertz (Hz)
- The unit meaning cycles per second.
- Core
- An independent processing unit within a CPU that can execute instructions.
- Multi-core processor
- A CPU containing two or more processing cores.
- Cache
- Small, fast memory close to or inside the CPU that stores frequently or recently used instructions and data.
- Cache hit
- A request for an item that is found in cache.
- Cache miss
- A request for an item that is not in cache and must be obtained from slower memory.
- Instruction set
- The complete collection of machine-code operations a particular processor can execute.
- Opcode
- The binary part of an instruction specifying the operation.
- Architecture
- The design and organisation of a processor, including its registers and instruction set.
- Parallel processing
- Carrying out more than one part of a task at the same time when the task and software permit it.
Clock Speed
A processor is controlled by electronic timing pulses produced by a clock. Clock speed is the number of cycles per second and is measured in hertz. A speed of 3 gigahertz represents approximately three billion clock cycles per second. Many processor operations require one or more clock cycles.
A higher clock speed can allow more instruction-processing steps to occur in a given time, so it can improve performance when other factors are similar. However, clock speed alone does not describe the amount of useful work completed per cycle. Different processor designs may execute instructions differently, and an application may spend time waiting for memory, storage or input/output.
Higher clock speed can also increase electrical power use and heat production. A processor may reduce its operating speed when it becomes too hot. Therefore, an examination answer should say that increasing clock speed generally can improve performance, not that it always makes every computer exactly proportionally faster.
Number Of Cores
A processor core can fetch, decode and execute instructions. A multi-core processor contains more than one core, allowing instructions from different tasks or different parts of one task to be processed at the same time. This can improve responsiveness when several applications are active.
The improvement depends on whether software can divide work into parallel sections. Some tasks have stages that depend on earlier results and therefore cannot be shared efficiently across many cores. The operating system and application must also be designed to schedule work effectively.
Doubling the number of cores does not necessarily double performance. Communication between cores, access to shared memory and parts of a program that remain sequential can limit the benefit.
Cache Size
The CPU can operate much faster than main memory. Cache reduces this difference by holding copies of instructions and data that the processor is likely to need. Accessing cache is faster than accessing RAM, so a cache hit can reduce waiting time.
A larger cache can hold more useful items and may increase the chance of a cache hit. It can therefore improve performance, especially when a program repeatedly uses the same instructions or nearby data. Cache is more expensive per unit of storage than RAM, so it is much smaller than main memory.
A larger cache does not guarantee a large improvement. If a program accesses a very large and unpredictable set of data, many requests may still be cache misses.
Instruction Sets And Processor Compatibility
An instruction set is the collection of operations that a processor can execute directly. It includes instructions for data movement, arithmetic, logic, comparison and control flow. Each machine-code instruction uses an opcode and may contain an operand or address information.
A program must be translated into machine code that matches the target processor’s instruction set. This is why software compiled for one processor architecture may need a different version for another architecture. The same high-level source code can often be translated separately for different processors.
Instruction-set design can affect performance because processors differ in the operations they support and the way instructions are encoded and executed. At this level, students are expected to understand the purpose of an instruction set rather than compare advanced instruction-set families.
Factors Affecting CPU Performance
| Factor | How An Increase May Help | Why Improvement May Be Limited |
|---|---|---|
| Clock speed | More processing cycles can occur each second. | Different architectures do different work per cycle; heat and waiting time can limit gains. |
| Number of cores | More tasks or parallel sections can be processed at the same time. | Software may not divide efficiently and some stages remain sequential. |
| Cache size | More frequently used instructions and data may be available quickly. | Programs with unpredictable access patterns may still have many cache misses. |
Typical Instruction Categories
| Category | Purpose | Illustrative Operation |
|---|---|---|
| Data movement | Copies a value between registers or memory. | Load a value from memory into a register. |
| Arithmetic | Performs a numerical calculation. | Add two values. |
| Logic | Performs a Boolean operation. | AND two binary values. |
| Comparison | Compares values and records a condition. | Check whether two values are equal. |
| Control flow | Changes the order of execution. | Jump to another instruction if a condition is true. |
Worked Examples
Comparing Two Processor Specifications
Question: Processor A has four cores at 3.6 GHz and Processor B has eight cores at 3.0 GHz. Explain why the faster choice cannot be identified from these figures alone.
- Clock speed suggests how many cycles each core completes per second.
- Core count suggests how much work may be processed in parallel.
- The application may or may not use eight cores effectively.
- Cache, processor design, memory access and heat also affect performance.
Answer: Neither specification alone proves which processor is faster; performance depends on the workload, software and other architectural factors.
Explaining A Cache Benefit
Question: A loop repeatedly uses the same instructions and small set of values. Explain why cache can improve execution speed.
- The instructions and values are likely to be copied into cache.
- Later requests are likely to be cache hits.
- The CPU can obtain the items from fast cache rather than slower RAM.
- The processor spends less time waiting for memory.
Answer: Repeatedly used instructions and values can remain in cache, reducing slower main-memory accesses.
Examination Guidance
- For every performance factor, explain both the mechanism and the limitation.
- Do not compare clock speeds from very different architectures as though frequency were the only factor.
- State that multiple cores help only when tasks can be processed concurrently or software supports parallel work.
- Define cache as fast memory holding frequently or recently used instructions and data.
- Define an instruction set as all machine-code operations supported by the processor.
Common Mistakes
- Claiming that twice the clock speed always gives twice the application speed.
- Claiming that every program automatically uses all available cores.
- Confusing cache with permanent secondary storage.
- Saying cache stores every file on the computer.
- Defining an instruction set as a user’s sequence of steps rather than processor machine-code operations.
Knowledge Check
1. How can a higher clock speed improve performance?
2. Why may additional cores not give proportional improvement?
3. How does cache improve performance?
4. What is an instruction set?
5. Why may software need different compiled versions for different processors?
