Moving a method from struct impl to trait causes performance degradation #69593
Description
Hi all, thanks for all the work on the rust project!
I tried this code:
use std::time::Instant;
fn main() {
let finder = OddFinderOne;
let start_time = Instant::now();
let result = find_nth_odd(finder, 1_000_000_000);
let elapsed_time = start_time.elapsed().as_millis();
println!("{} in {} ms", result, elapsed_time);
}
fn find_nth_odd(odd_finder: OddFinderOne, n: u64) -> u64 {
let mut i = 0;
let mut odd_count = 0;
while odd_count != n {
i += 1;
if odd_finder.is_odd(i) {
odd_count += 1;
}
}
i
}
trait OddFinder {
fn is_odd(&self, n: u64) -> bool;
}
struct OddFinderOne;
impl OddFinder for OddFinderOne {
fn is_odd(&self, n: u64) -> bool {
n % 2 == 1
}
}I expected it to perform similarly to this code:
use std::time::Instant;
fn main() {
let finder = OddFinder;
let start_time = Instant::now();
let result = find_nth_odd(finder, 1_000_000_000);
let elapsed_time = start_time.elapsed().as_millis();
println!("{} in {} ms", result, elapsed_time);
}
fn find_nth_odd(odd_finder: OddFinder, n: u64) -> u64 {
let mut i = 0;
let mut odd_count = 0;
while odd_count != n {
i += 1;
if odd_finder.is_odd(i) {
odd_count += 1;
}
}
i
}
struct OddFinder;
impl OddFinder {
fn is_odd(&self, n: u64) -> bool {
n % 2 == 1
}
}Instead, this happened:
The version with the is_odd method defined directly on the struct completes in 1032 ms, while the version where is_odd is defined as part of the trait completes in 1340 ms. I would expect this abstraction to be zero cost, since the find_nth_odd method in both cases is defined as taking the concrete struct as an argument (as opposed to a trait object).
Meta
rustc --version --verbose:
rustc 1.41.0 (5e1a79984 2020-01-27)
binary: rustc
commit-hash: 5e1a799842ba6ed4a57e91f7ab9435947482f7d8
commit-date: 2020-01-27
host: x86_64-unknown-linux-gnu
release: 1.41.0
LLVM version: 9.0
rustc +nightly --version --verbose
rustc 1.43.0-nightly (0eb878d2a 2020-02-28)
binary: rustc
commit-hash: 0eb878d2aa6e3a1cb315f3f328681b26bb4bffdb
commit-date: 2020-02-28
host: x86_64-unknown-linux-gnu
release: 1.43.0-nightly
LLVM version: 9.0
I see the same behavior with both the latest stable and nightly compilers. I run both version of the program in release mode. The reported performance numbers have proven to be very repeatable.
Let me know if I can provide any additional useful information. You can find my repo containing this benchmark code at the link below.