Write a bench that checks code and valid of a 4-input priority encoder on every pattern, and catches eight planted faults.
Hardware context: An interrupt line shared by four sensors, and a box of old priority encoders to sort. Write the bench that catches every planted fault in a priority encoder.
A team has eight versions of a 4-input priority encoder from an old project. Some are right and some are not, and their old bench only printed. You must write a bench that sorts them.
This is a Verify problem. You write the bench, not the encoder. Your bench drives the inputs r and reads code and valid. It adds 1 to errors for every input pattern where an output is wrong.
The check runs your bench on one correct encoder and on eight faulty ones. Several faults show on a single pattern only. You pass when errors stays 0 on the correct encoder and ends above 0 on every faulty one.
Signals in the Timeline
| Signal | Width | Role |
|---|---|---|
variant | 4 | input of each check |
Full specification
module bench (
output logic [3:0] r, // you drive the four inputs of the encoder
input logic [1:0] code, // its outputs
input logic valid,
output int errors // add 1 for every input pattern where an output is wrong
);
// your initial block
endmodule
| Rule | Detail |
|---|---|
| The encoder | code is the number of the highest active input of r; valid is 1 when any input is active. |
| No input active | code = 0 and valid = 0. |
| Timing | The encoder is plain gates. Wait 1 ns after you change r before you read the outputs. |
| Time limit | Your bench must finish within 35 ns. |
| errors | Start at 0. Add 1 for each wrong pattern. Never count down. |
For each encoder you get alarm = yes or no. The correct encoder must give no alarm. Each faulty encoder must give an alarm.
Examples
Example 1
The Correct Encoder Must Not Raise an Alarm- Input
Encoder: the correct design
- Output
alarm = no
On the correct encoder every pattern matches, so errors must still be 0 at the end.
Example 2
An Encoder Where the Lowest Input Wins- Input
Encoder: Lowest input wins (with two or more inputs, the lowest one is encoded)
- Output
alarm = yes
It agrees with the rule whenever one input is active. Only patterns with two or more inputs show the fault.
Example 3
An Encoder Whose Valid Is Stuck at 1- Input
Encoder: valid stuck at 1 (valid is 1 even when no input is active)
- Output
alarm = yes
Every code is right. Only the pattern 0000, read through valid, shows the fault.
Constraints and Contract
- Use only your ports. Names from the fixture, $root and instance names (%m) are refused before the run.
- Drive r from one initial block, and finish within the time limit.
Hints
Hint 1
Four inputs have sixteen patterns. A loop from 0 to 15 visits them all, and a fault that shows on one pattern cannot hide.
Hint 2
Work out the expected code yourself. Walk the bits of r from 0 to 3 and keep the last one that is set. Do not forget 0000.
Hint 3
Check both outputs. A fault in valid alone leaves every code right.
What the grader checks: Your bench runs on one correct encoder and on eight encoders with one planted fault each. Six of the faults show on a single input pattern. Passing shows that your bench checks both outputs on every pattern and raises no false alarm.
9 checks on Icarus · browser samples / server checks. Run the examples first, then check the full regression. Full checks use the server judge. Downloads contain public samples only. A passing result establishes the tested cases, not a proof for all possible inputs.
See all 2 steps and the final capstone