Round a signed Q8.4 number to the nearest whole number, sending every tie to the even neighbour.
Hardware context: A fixed-point gain stage of an audio mixer and the display rounding of a digital scale. Round a signed Q8.4 number to the nearest whole number with ties to even.
A digital scale averages many readings and shows a whole number of grams. Its first version rounded every half upward. Over a day of weighing, the totals came out too high.
This is an Implement problem. Write round_even. The input x is a signed Q8.4 number. The output y is the nearest whole number. When x is exactly half way between two whole numbers, y is the even one of the two.
The rule is the same for negative values: -2.5 rounds to -2 and -3.5 rounds to -4.
The check applies eight values of x, one per nanosecond, and records y after each.
Signals in the Timeline
| Signal | Width | Role |
|---|---|---|
script | 96 | input of each check |
Full specification
module round_even (
input logic signed [11:0] x, // signed Q8.4: -128.0 to 127.9375
output logic signed [8:0] y // x rounded to a whole number, ties to even
);
// your design
endmodule
| Fraction part of x, above the whole number below it | y |
|---|---|
| Less than one half | The whole number below x. |
| More than one half | The whole number above x. |
| Exactly one half | The even one of the two. |
The output has nine bits because the largest tie, 127.5, rounds up to 128.
The result fields are named y_step0 to y_step7, shown as signed decimal numbers.
Examples
Example 1
Values with No Tie- Input
x = 1.25; x = 1.75; x = 2.0625; x = 2.9375; x = 0.4375; x = 0.5625; x = 10; x = 99.875
- Output
y_step0 = 1, y_step1 = 2, y_step2 = 2, y_step3 = 3, y_step4 = 0, y_step5 = 1, y_step6 = 10, y_step7 = 100
Each value goes to its nearest whole number: 1.25 gives 1 and 1.75 gives 2. No value here is exactly half way, so every rounding rule agrees.
Example 2
Positive Ties- Input
x = 0.5; x = 1.5; x = 2.5; x = 3.5; x = 4.5; x = 5.5; x = 126.5; x = 127.5
- Output
y_step0 = 0, y_step1 = 2, y_step2 = 2, y_step3 = 4, y_step4 = 4, y_step5 = 6, y_step6 = 126, y_step7 = 128
Each value is exactly half way. 0.5 gives 0 and 1.5 gives 2: the even neighbour each time. 127.5 gives 128, which is why y has nine bits.
Example 3
Negative Values with No Tie- Input
x = -1.25; x = -1.75; x = -2.0625; x = -2.9375; x = -0.4375; x = -0.5625; x = -10; x = -99.875
- Output
y_step0 = -1, y_step1 = -2, y_step2 = -2, y_step3 = -3, y_step4 = 0, y_step5 = -1, y_step6 = -10, y_step7 = -100
-1.25 gives -1 and -1.75 gives -2. The nearest whole number is found the same way on both sides of zero.
Constraints and Contract
- Inputs are known 0/1 values. Each value is held for one nanosecond and y is sampled at its end.
- Keep the module name and the ports. The design is combinational.
Hints
Hint 1
Split x into its top eight bits and its low four bits. In two's complement the top bits are the whole number at or below x, even when x is negative.
Hint 2
The low four bits count sixteenths above that whole number, from 0 to 15. One half is eight sixteenths.
Hint 3
There are three cases: above eight, below eight and exactly eight. In the last case, look at the lowest bit of the whole number to see if it is odd.
What the grader checks: Seven directed and eight seeded cases of eight steps each run on the block. Every output is compared with a model written in JavaScript. Other rounding modes are outside this problem.
15 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