Is there any difference between the two ways to round a number to a multiple of 8 in C

Question

If size is a int variable

int size = 10;

And I want to round up size to a multiple of 8, what is the difference between the two ways:

A:

size = 1 + ((size - 1)/8);
size = size * 8;

B:

size = (size/8+1)*8;

Thanks in advance.

Are both working? Then semantically there is no difference. What is more efficient and readable? You can guess. — Eugene Sh., Sep 26 '16 at 18:04
Wow. Three answers saying essentially the same thing within 14s of each other. — John Bollinger, Sep 26 '16 at 18:06
Note that method A gives an incorrect result if `size` is zero. — nwellnhof, Sep 26 '16 at 18:19
At the very least, you should review [Rounding integer division instead of truncating](http://stackoverflow.com/questions/2422712/). I'd like to be convinced this isn't just a duplicate of that. — Jonathan Leffler, Sep 26 '16 at 19:58

score 3 · Accepted Answer · answered Sep 26 '16 at 18:05

Is there any difference between [these] two ways to round a number to a multiple of 8 in C[?]

Yes. They produce different results for inputs that are already multiples of 8. Approach (A) returns such inputs unchanged, but approach (B) returns the next larger multiple of 8 (supposing no overflow).

score 2 · Answer 2 · answered Sep 26 '16 at 18:05

The two are doing different things:

The first sequence correctly keeps multiples of 8 in place
The second sequence rounds 8 up to 16, 16 up to 24, and so on.

Here is the sequence of outputs for both snippets for values 0..19, with differences highlighted:

0 8 8
1 8 8
2 8 8
3 8 8
4 8 8
5 8 8
6 8 8
7 8 8
8 8 16
9 16 16
10 16 16
11 16 16
12 16 16
13 16 16
14 16 16
15 16 16
16 16 24
17 24 24
18 24 24
19 24 24

score 2 · Answer 3 · answered Sep 26 '16 at 18:05

2

Method A keeps multiples of 8 the same value and rounds other values up.

Method B won't work for multiples of 8, as it rounds those up as well.

For example: (16/8+1)*8 == (2+1)*8 == 3*8 == 24.

answered Sep 26 '16 at 18:05

dbush

205,898
23
218
273

score 1 · Answer 4 · answered Sep 26 '16 at 19:00

As others has stated, the A and B behave differently. Just try with size = 8 for both. Solution B is for sure wrong as 8 results in 16.

Rounding can be done in several ways. Especially negative numbers can be tricky. Should -4 result in -8 or 0?

I would go for round to nearste with Round half away from zero like this:

size = (size >= 0) ? 8*((size+4)/8) : 8*((size-4)/8);

See https://en.wikipedia.org/wiki/Rounding for more about rounding.

score 1 · Answer 5 · answered Sep 26 '16 at 19:15

1

Rounding up to a power of 2 can be done like

size = (size + 7) & (~7)

Sometimes helps when you're on a CPU that has no fast multiplication / division

answered Sep 26 '16 at 19:15

tofro

5,640
14
31

1

7 isn't a power of two. You should explain a bit better (maybe if N is a power of two, then `(size + N - 1) & ~(N - 1)`…). – Jonathan Leffler Sep 26 '16 at 20:00

score 1 · Answer 6 · answered Sep 26 '16 at 19:19

Different results when size is 8, 16, 24, 32 ...
Different results when size is -7, -15, -23 ....

A is correct for values greater than 1.
B is not correct for positive multiples of 8.

Both fail for size == 0. Unclear if size < 0 is important for OP.

To do unsigned rounding up to a multiple of N which also works for size == 0.

#define N 8
unsigned size;
unsigned rounded_size = (size + N - 1) % N;

Is there any difference between the two ways to round a number to a multiple of 8 in C

6 Answers6