Computer hardware and operation

A1.1.1 Inside the CPU

The central processing unit does most of the processing in a device. Three groups of parts work together:

Registers

Buses

Cores

Next: the GPU →

A1.1.2 The role of the GPU

A graphics processing unit speeds up work by doing a huge number of simple calculations at the same time. This is called parallel processing. It was built for graphics, and its structure turned out to suit other kinds of work too.

Next: memory →

A1.1.4 Primary memory

Primary memory is memory the CPU uses directly. Each type exists because no single memory can be fast, large, cheap and permanent all at once.

The CPU looks in cache first, then RAM, then secondary storage. Finding the data in cache is a cache hit. Not finding it is a cache miss, and the CPU has to wait for slower memory.

Next: the instruction cycle →

A1.1.5 Fetch, decode, execute

Every instruction a CPU runs goes through the same three stages:

1. Fetch: the instruction at the address in the PC is copied from memory into the IR. The PC moves on.
2. Decode: the control unit works out what the instruction means.
3. Execute: the CPU carries it out, which might use the ALU, memory or a device.

The Little Man Computer (LMC) is a teaching model of a CPU. Each instruction is a three-digit code, so you can watch the registers change at every stage.

Next: storage →

A1.1.7 Secondary storage

Secondary storage keeps data with the power off. Choosing one is a trade-off between speed, capacity, cost and portability.

Internal

M.2 SSDs are small sticks that slot into the motherboard. eMMC is flash storage soldered to the board in low-cost phones and laptops: slower than an SSD, but enough for basic use.

External

Next: compression →

A1.1.8 Compression

Compression stores information in fewer bits, so files take less space and move faster across a network.

Run-length encoding (lossless)

RLE replaces each run of repeated characters with a count and the character. It is excellent for data with long runs and poor for data without them, and it can even make a file bigger.

Transform coding (lossy)

JPEG and MP3 both use transform coding. For a JPEG:

1. Split the image into small square blocks.
2. Transform each block with the discrete cosine transform (DCT), which turns pixel values into low- and high-frequency coefficients.
3. Quantize: reduce the precision of the high-frequency coefficients. Eyes notice this fine detail least. This is the step that loses data.
4. Encode the result with a zigzag scan, then RLE, then Huffman coding.

def compress(text):
    if not text:
        return ""
    result, count = "", 1
    for i in range(1, len(text)):
        if text[i] == text[i - 1]:
            count += 1
        else:
            result += str(count) + text[i - 1]
            count = 1
    return result + str(count) + text[-1]


def decompress(code):
    result, digits = "", ""
    for ch in code:
        if ch.isdigit():
            digits += ch
        else:
            result += ch * int(digits)
            digits = ""
    return result


print(compress("MMMMMOOPPPPPPPQ"))   # 5M2O7P1Q
print(decompress("5M2O7P1Q"))        # MMMMMOOPPPPPPPQ

Next: cloud services →

A1.1.9 Cloud service types

The three models differ in how much you manage yourself and how much the provider manages for you.

Next: review →

Quick quiz

Exam-style questions

These are written for this site in the style of Paper 1. Answer on paper first, then open the mark points.

1. Describe the function of the control unit. [2]

   Mark points

   • Directs the fetch, decode and execute of instructions [1].
   • Sends control signals to coordinate memory, the ALU and I/O devices [1].

2. Outline why the memory address register is needed during a fetch. [2]

   Mark points

   • It holds the address of the location being accessed [1].
   • It places that address on the address bus so memory can return the contents [1].

3. A video-editing workstation has a choice of a dual-core or an eight-core CPU at the same clock speed. Explain one reason the eight-core CPU may not run an older program faster. [3]

   Mark points

   • The program may have been written to use a single thread [1].
   • It can only run on one core at a time [1].
   • So the extra cores stay idle for that program and its speed does not change [1].

4. Explain why a computer has cache memory as well as RAM. [3]

   Mark points

   • RAM is much slower than the CPU [1].
   • Cache keeps frequently used data and instructions close to the CPU [1].
   • More requests are cache hits, so the CPU spends less time waiting [1].

5. Describe the role of the buses during the fetch stage. [4]

   Mark points

   • The address in the MAR is placed on the address bus [1].
   • The control bus carries a read signal [1].
   • Memory places the instruction on the data bus [1].
   • The instruction arrives in the MDR and is copied to the IR [1].

6. A school keeps 400 TB of old CCTV footage that is rarely viewed. Suggest whether it should use HDDs or SSDs, and justify your choice. [3]

   Mark points

   • HDDs [1].
   • They cost less per GB, which matters at this volume [1].
   • Read speed matters little because the footage is rarely viewed [1].

7. Distinguish between lossless and lossy compression, giving a suitable use of each. [4]

   Mark points

   • Lossless restores the original exactly [1]; for example, a source code or text file [1].
   • Lossy permanently removes data that people are less likely to notice [1]; for example, a JPEG photo or streamed audio [1].

8. A start-up is choosing IaaS rather than buying servers. Outline one advantage and one disadvantage, and state one thing it remains responsible for. [3]

   Mark points

   • Advantage: low upfront cost, or it can scale up as users grow [1].
   • Disadvantage: it needs the technical skill to run the systems [1].
   • Responsible for: its operating systems, applications or data security [1].