iter_mut is slightly slower than code with a manual for loop #144118
Description
Code using iter_mut is slightly slower than code with a manual for loop. Today's nightly results:
$ cargo bench
Finished `bench` profile [optimized] target(s) in 0.01s
Running unittests src/lib.rs (target/release/deps/bench_round_up-6e3523a74735e3c8)
running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
Running benches/bench.rs (target/release/deps/bench-f02e501c16b73b8b)
running 2 tests
test bench_round_up_for ... bench: 730,833.00 ns/iter (+/- 1,163.95)
test bench_round_up_iter ... bench: 740,568.00 ns/iter (+/- 1,857.11)
test result: ok. 0 passed; 0 failed; 0 ignored; 2 measured; 0 filtered out; finished in 0.88s
The functions:
pub fn round_up_iter(d: &mut [u8]) -> Option<u8> {
match d.iter().rposition(|&c| c != b'9') {
Some(i) => {
// d[i+1..n] is all nines
d[i] += 1;
d.iter_mut().skip(i + 1).for_each(|c| *c = b'0');
None
}
None if d.is_empty() => {
// an empty buffer rounds up (a bit strange but reasonable)
Some(b'1')
}
None => {
// 999..999 rounds to 1000..000 with an increased exponent
d[0] = b'1';
d.iter_mut().skip(1).for_each(|c| *c = b'0');
Some(b'0')
}
}
}
pub fn round_up_for(d: &mut [u8]) -> Option<u8> {
match d.iter().rposition(|&c| c != b'9') {
Some(i) => {
// d[i+1..n] is all nines
d[i] += 1;
for j in i + 1..d.len() {
d[j] = b'0';
}
None
}
None if d.len() > 0 => {
// 999..999 rounds to 1000..000 with an increased exponent
d[0] = b'1';
for j in 1..d.len() {
d[j] = b'0';
}
Some(b'0')
}
None => {
// an empty buffer rounds up (a bit strange but reasonable)
Some(b'1')
}
}
}This holds even if the order of match arms is the same.