Final Test: Computer Architecture

The grade formula is as follows:

Grade = 0.4 * Theory + 0.6 * Programming

Part 1. Theoretical Questions

Answer the theoretical questions in a few sentences (1-2 paragraphs for a question).

Use your own words. Copy-paste from external sources and other students will result in 100% penalty.

1. Describe how integers are represented in the binary format (unsigned integers, 2’s complement). What is the difference between integer sign- and zero-extension? What is integer overflow? Explain the floating-point format (sign, exponent, fraction). Describe single and double precision. What is implicit 1. in fraction? Why exponents are biased (and what is bias)? What are floating-point overflow and underflow?

2. Describe the concept of pipeline. List the 5 stages and give brief descriptions for them. What pipeline hazards are? List the types of hazards and the ways to prevent them (with brief definitions). Give an example of a hazard situation and how it can be handled. What is branch prediction is needed for? How the loop unrolling optimisation can help improve pipeline performance?

3. Give a brief definition of a cache. Why do we need it? Explain how caching works (block, index, tag, valid bit, dirty bit). Give the definition of associativity (direct-mapped, set associative, fully associative). What is the difference between write-through and write-back? What is replacement policy (what types of policy do you know)? How many cache levels are typical for modern processors? What problem can caches create for multicore processors?

Part 2. Programming Task

Write a RISC-V assembly program that inputs integer value x and calculates the value of the f(x) mathematical function according to the specified equations.

Requirements:

• x is assumed to be a non-negative value from 0 to 12.
• f(x) must be implemented as a single function and must comply with RISC-V calling conventions.
• f(x) must use callee-saved registers (s0, s1, etc.) to store intermediate results of calculations (these registers are saved to the stack and restored from the stack by f(x)).

The f(x) function is as follows (this is an example, find your variant below):

```
f(x) = -2 * x + 4 if x < 5
f(x) = -4**x - 4 if x >= 5
f(x) = 5 * x**7 - 2 * x**4 - 2 * x if x == 7
f(x) = -7 if x > 9
```

Evaluation criteria:

1. Correct program structure (input x, print f(x), switch between cases of x) costs 1 point.
2. Each of the f(x) cases costs 1.5 points (must give a valid result).
3. Correct function implementation consts 1.5 point (call and result, passing argument and result).
4. Correct work with the stack costs 1.5 points (correct saving and loading registers).

Variants

Choose your variant number according to your number in your group.

See the lists of students here.

See the variants here.

Here the retake the variants here.