GCSE Computer Science Paper 3
Covers CPU and the Fetch-Execute Cycle, Memory and Storage, Networks, Topologies and Protocols and 9 more.
Questions
Question 1 [2 marks]
Representing Text, Images, Sound and Compression
An image uses a colour depth of 4 bits per pixel.
Calculate the maximum number of different colours that can be represented using 4 bits per pixel, showing your method.
Question 2 [4 marks]
Algorithm Design and Trace Tables
The following pseudocode is run:
total = 0
for i = 1 to 5
if i mod 2 == 0 then
total = total + i
endif
next i
print total
State what is output by this pseudocode, explaining which values of i are added to total.
Question 3 [4 marks]
Ethics, Law and Environmental Impact
The UK GDPR sets out principles for how personal data must be handled.
State four principles that organisations must follow when processing personal data under the UK GDPR.
Question 4 [4 marks]
Cyber Security Threats and Measures
A teacher is explaining different types of cyber security threat to a class, other than malware and phishing.
State four other types of cyber security threat.
Question 5 [2 marks]
CPU and the Fetch-Execute Cycle
A quad-core CPU has four cores, each with the same clock speed as an otherwise identical single-core CPU.
Explain why running an application that has been written to use only one core may not be much faster on the quad-core CPU than on the single-core CPU.
Question 6 [2 marks]
Number Systems: Binary and Hexadecimal
Subtract the 8-bit binary number 00011010 from the 8-bit binary number 01000101, giving your answer as an 8-bit binary number.
Question 7 [2 marks]
Boolean Logic and Logic Gates
A logic circuit has two inputs, A and B, and its truth table is: A=0,B=0 gives output 0 ; A=0,B=1 gives output 1 ; A=1,B=0 gives output 1 ; A=1,B=1 gives output 0.
State the name of the single logic gate that matches this truth table, and write the Boolean expression for its output using A and B.
Question 8 [3 marks]
Networks, Topologies and Protocols
A student notices that their bank's website address begins with https rather than http.
Explain what the 's' in https indicates, and explain why this is important when entering banking details online.
Question 9 [4 marks]
Memory and Storage
A laptop's RAM becomes full while several programs are open at once.
Explain how virtual memory allows the laptop to keep running, and state one disadvantage of relying on virtual memory.
Question 10 [4 marks]
Boolean Logic and Logic Gates
A NAND gate has two inputs, C and D, and its output is the opposite of what an AND gate would give for the same inputs.
Write a Boolean expression for the NAND gate's output using C and D, and state its output when C=1 and D=1.
Question 11 [5 marks]
Searching and Sorting Algorithms
A merge sort and a bubble sort are both used to sort the same list of 1000 already-sorted (ascending) items.
Discuss whether a bubble sort would perform noticeably better on this already-sorted list than on a randomly ordered list of the same size, and whether the same is true for merge sort.
Discuss both fixed-pass and no-swap early-exit bubble sort. For merge sort assume the standard divide-to-singletons implementation, and note how adaptive variants could change the result.
Question 12 [5 marks]
Programming Constructs
A programmer declares a variable inside a subroutine, and a different variable outside any subroutine, in the main part of the program.
Explain where local and global variables are declared and where each is accessible. Then explain one risk of choosing a global for temporary subroutine data.
Question 13 [4 marks]
Number Systems: Binary and Hexadecimal
A student claims: 'Converting between hexadecimal and binary always takes longer than converting between denary and binary, since hexadecimal is a completely different, unrelated number system.'
Discuss whether this claim is accurate, referring to how a hexadecimal digit relates to a group of binary bits.
Question 14 [5 marks]
Representing Text, Images, Sound and Compression
A simple black-and-white fax image contains large areas of a single colour, for example long runs of white pixels followed by long runs of black pixels.
Explain how run-length encoding (RLE) could compress this image, and discuss whether RLE would work as well on a typical photograph of a busy street scene.
Question 15 [5 marks]
Operating Systems and Utility Software
A school is deciding whether to install operating system security updates on all of its computers automatically overnight, or to have an IT technician install them manually during the day, one computer at a time.
Discuss the advantages and disadvantages of each approach.
Question 16 [5 marks]
Representing Text, Images, Sound and Compression
A film studio must choose between storing its finished film using lossless compression or lossy compression, given the film will later be re-edited and re-compressed several times before its final release.
Discuss why lossless compression is more appropriate than lossy compression for this stage of the film's production, even though it produces larger files.
Model solutions
| Question 1[2 marks] | |
|---|---|
| Answer or working | Marks |
| Uses 2^4 to count the available bit patterns. | M1 |
| 16 colours; correct method and answer earn both marks. | A1 |
| Final answer: 16 colours | |
| Question 2[4 marks] | |
|---|---|
| Answer or working | Marks |
| the loop runs with i taking the values 1, 2, 3, 4 and 5 in turn | B1 |
| the if statement only adds i to total when i mod 2 equals 0, meaning i is even | B1 |
| so only i = 2 and i = 4 are added to total, giving 2 + 4 = 6 | B1 |
| the output is 6 | A1 |
| Final answer: 6 (only the even values, 2 and 4, are added) | |
| Question 3[4 marks] | |
|---|---|
| Answer or working | Marks |
| one distinct UK GDPR principle from lawfulness/fairness/transparency, purpose limitation, data minimisation, accuracy, storage limitation, integrity/confidentiality, accountability | B1 |
| a second different principle from that set | B1 |
| a third different principle from that set | B1 |
| a fourth different principle from that set; do not count components of one grouped principle twice | B1 |
| Final answer: Any four distinct principles: lawfulness, fairness and transparency (one grouped principle); purpose limitation; data minimisation; accuracy; storage limitation; integrity and confidentiality; accountability. | |
| Question 4[4 marks] | |
|---|---|
| Answer or working | Marks |
| one distinct relevant threat other than malware/phishing, such as brute force, DoS/DDoS, SQL injection, shoulder surfing, pharming or interception | B1 |
| a second different valid threat from that range or another sound example | B1 |
| a third different valid threat | B1 |
| a fourth different valid threat; do not credit DoS and DDoS separately as the same underlying denial-of-service threat | B1 |
| Final answer: Brute-force password attacks, denial of service, SQL injection and interception/man-in-the-middle are four threats other than malware/phishing. Data theft, social engineering other than phishing or pharming may also be valid distinct threats. | |
| Question 5[2 marks] | |
|---|---|
| Answer or working | Marks |
| the application cannot split its sequential work across the extra cores, so only one core executes its instructions at a time | B1 |
| that core has the same clock speed as the single-core CPU, so application execution is similar; other cores can help background tasks but do not give a fourfold application speedup | B1 |
| Final answer: Only one core executes this application at a time. With the same per-core speed, the other three cores cannot substantially accelerate its sequential work, although they can run background tasks. | |
| Question 6[2 marks] | |
|---|---|
| Answer or working | Marks |
| Valid subtraction method: binary borrowing, adding the two's complement, or 69-26=43 followed by conversion. | M1 |
| 00101011; accept a correct answer without working for both marks. | A1 |
| Final answer: 00101011 | |
| Question 7[2 marks] | |
|---|---|
| Answer or working | Marks |
| Names XOR (exclusive OR). | B1 |
| Gives A XOR B, or (A AND NOT B) OR (NOT A AND B), or another equivalent Boolean expression. | B1 |
| Final answer: XOR gate; output = A XOR B | |
| Question 8[3 marks] | |
|---|---|
| Answer or working | Marks |
| the s denotes secure HTTP using TLS encryption | B1 |
| HTTP without transport encryption can expose plaintext data to interception | B1 |
| encrypting the browser-server connection prevents an interceptor reading banking credentials without the relevant keys | B1 |
| Final answer: The s denotes secure HTTP: TLS encrypts traffic between browser and server so intercepted banking credentials are not readable as plaintext. | |
| Question 9[4 marks] | |
|---|---|
| Answer or working | Marks |
| virtual memory uses space on secondary storage (such as the hard disk) as if it were extra RAM | B1 |
| the operating system moves data between RAM and this secondary storage space as it is needed | B1 |
| this allows more programs to keep running than would fit in the physical RAM alone | B1 |
| disadvantage: secondary storage is much slower than RAM, so relying on virtual memory slows the computer down | B1 |
| Final answer: Virtual memory swaps data between RAM and secondary storage so more can run at once, but access is much slower | |
| Question 10[4 marks] | |
|---|---|
| Answer or working | Marks |
| Combines C and D with AND, or uses an equivalent De Morgan form. | M1 |
| Negates the whole conjunction: NOT (C AND D), accepting NOT C OR NOT D. | M1 |
| Evaluates C AND D as 1 for C=D=1, or correctly substitutes into an equivalent expression. | M1 |
| Gives output 0; full evaluation credit for the correct stated output. | A1 |
| Final answer: output = NOT (C AND D); with C=1, D=1 the output is 0 | |
| Question 11[5 marks] | |
|---|---|
| Answer or working | Marks |
| Early-exit bubble detects a pass with no swaps | B1 |
| Sorted input needs 999 adjacent comparisons and no swaps | B1 |
| Random input generally requires many passes and up to quadratic work | B1 |
| Fixed-pass bubble does not gain the same reduction in comparisons | B1 |
| Standard merge retains O(n log n) division and merge work despite sorted input | B1 |
| Final answer: An early-exit bubble sort compares all 999 adjacent pairs once, makes no swaps and terminates, giving O(n) work on this sorted input. On random input it usually needs many passes and can take O(n^2). A fixed-pass bubble sort still makes its prescribed comparisons, though it avoids swaps. Standard merge sort still divides and merges O(log n) levels with O(n) work per level, so remains O(n log n); adapted merge implementations may exploit existing runs. | |
| Question 12[5 marks] | |
|---|---|
| Answer or working | Marks |
| Local is declared inside the subroutine | B1 |
| Local is accessible only within its scope | B1 |
| Global is declared at program scope | B1 |
| Global can be accessible across subroutines subject to language rules | B1 |
| Local temporary state reduces accidental interference | B1 |
| Final answer: A local variable is declared inside a subroutine and is accessible only within that scope. A global variable is declared in an enclosing program-level scope and can be accessed by subroutines where the language scope rules allow it. Keeping temporary values local avoids accidental interference; a separate local binding with the same name may hide a global without changing it. | |
| Question 13[4 marks] | |
|---|---|
| Answer or working | Marks |
| Explains 16=2^4, so one hexadecimal digit corresponds to four binary bits. | B1 |
| Describes grouping binary bits in fours from the right, padding the leftmost group if needed, and mapping each group to hex (or reversing the process). | B1 |
| Gives a correct example, such as hexadecimal A = binary 1010. | B1 |
| Explains that denary has no direct four-bit grouping because 10 is not a power of 2, so direct hex conversion is often simpler; rejects the claim that hex is unrelated or always slower. | B1 |
| Final answer: The claim is false: each hex digit maps directly to four binary bits because 16=2^4, so conversion can use direct grouping and is often simpler than conversion via denary. | |
| Question 14[5 marks] | |
|---|---|
| Answer or working | Marks |
| RLE replaces a consecutive run with its value and repeat count; either ordering is acceptable. | B1 |
| Applies this to the fax, e.g. forty white pixels become a white-value/count-40 pair. | B1 |
| Long runs allow one pair to replace many pixels, substantially reducing data for the stated fax. | B1 |
| Busy photographs usually have frequent changes and few long identical runs. | B1 |
| Short runs give less benefit and count overhead may even enlarge the data, so RLE is generally less effective on the photograph. | B1 |
| Final answer: RLE compresses the fax well by replacing long runs of one colour with a value and count, but works poorly on a busy photograph, which has too few long runs of identical pixels to compress effectively | |
| Question 15[5 marks] | |
|---|---|
| Answer or working | Marks |
| overnight automation reduces staff effort and can keep the fleet patched consistently | B1 |
| overnight restarts/installation reduce disruption during lessons | B1 |
| unattended updates can cause compatibility/failure problems before a technician is present | B1 |
| manual staged installation allows checks before widespread deployment | B1 |
| manual one-at-a-time rollout consumes technician time and leaves some machines unpatched longer | B1 |
| Final answer: Overnight automation patches many machines promptly with little classroom disruption and less technician time, but failures/reboots or incompatible updates may affect many before review and powered-off devices may miss updates. Manual staged updates allow checks/control but take staff time, interrupt day use and delay protection. Test, schedule and monitor deployment with rollback options. | |
| Question 16[5 marks] | |
|---|---|
| Answer or working | Marks |
| Lossy compression irreversibly discards information from the input. | B1 |
| Repeated edits and lossy re-encoding can accumulate artefacts and reduce final quality; avoid claiming that every identical re-encoding must lose new information. | B1 |
| Lossless decoding restores the exact input data at each stage. | B1 |
| Larger lossless masters cost storage, transfer bandwidth and potentially processing time. | B1 |
| Justifies preserving a lossless editing master for later edits, with lossy compression if desired for final distribution. | B1 |
| Final answer: Lossless masters preserve exact data through repeated editing and encoding, avoiding the risk of cumulative lossy artefacts. Larger files require more storage and transfer resources; lossy compression can be reserved for distribution. | |