Skip to content

Add benchmarks for impl Debug for str #124551

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
May 10, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions library/core/benches/str.rs
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@ use test::{black_box, Bencher};

mod char_count;
mod corpora;
mod debug;
mod iter;

#[bench]
Expand Down
79 changes: 79 additions & 0 deletions library/core/benches/str/debug.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,79 @@
//! This primarily benchmarks `impl Debug for str`,
//! and it also explicitly tests that we minimizes calls to the underlying `Write`r.
//! While that is an implementation detail and there are no guarantees about it,
//! we should still try to minimize those calls over time rather than regress them.

use std::fmt::{self, Write};
use test::{black_box, Bencher};

#[derive(Default)]
struct CountingWriter {
buf: String,
write_calls: usize,
}

impl Write for CountingWriter {
fn write_str(&mut self, s: &str) -> fmt::Result {
self.buf.push_str(s);
self.write_calls += 1;
Ok(())
}
}

fn assert_fmt(s: &str, expected: &str, expected_write_calls: usize) {
let mut w = CountingWriter::default();

write!(&mut w, "{s:?}").unwrap();
assert_eq!(s.len(), 64);
assert_eq!(w.buf, expected);
assert_eq!(w.write_calls, expected_write_calls);
Comment on lines +27 to +29
Copy link
Member

@the8472 the8472 Apr 30, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Generally it's better to separate tests from benchmarks. That makes it easier to rewrite the benchmarks in case the optimizer gets too clever and optimizes things away. And it measures fewer things which should increase the SNR.
Right now this would also be benchmarking string comparison.

I know we have a few benchmarks which mix things, but they're not a good example to follow.

You can still write them in a single #[bench] function by running the asserts outside the benchmark loop.
Or you can declare the test strings as consts so they can be reused between tests and benches.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can still write them in a single #[bench] function by running the asserts outside the benchmark loop.
Or you can declare the test strings as consts so they can be reused between tests and benches.

I first tried to extract those to the tests directory, but then ended up leaving them together inside the #[bench] fn outside of the benchmark loop as you suggested.

IMO it makes sense to co-locate these, as especially counting / optimizing the write calls is done purely for performance reasons.

}

#[bench]
fn ascii_only(b: &mut Bencher) {
let s = "just a bit of ascii text that has no escapes. 64 bytes exactly!!";
assert_fmt(s, r#""just a bit of ascii text that has no escapes. 64 bytes exactly!!""#, 3);
b.iter(|| {
black_box(format!("{:?}", black_box(s)));
});
}

#[bench]
fn ascii_escapes(b: &mut Bencher) {
let s = "some\tmore\tascii\ttext\nthis time with some \"escapes\", also 64 byte";
assert_fmt(
s,
r#""some\tmore\tascii\ttext\nthis time with some \"escapes\", also 64 byte""#,
21,
);
b.iter(|| {
black_box(format!("{:?}", black_box(s)));
});
}

#[bench]
fn some_unicode(b: &mut Bencher) {
let s = "egy kis szöveg néhány unicode betűvel. legyen ez is 64 byte.";
assert_fmt(s, r#""egy kis szöveg néhány unicode betűvel. legyen ez is 64 byte.""#, 3);
b.iter(|| {
black_box(format!("{:?}", black_box(s)));
});
}

#[bench]
fn mostly_unicode(b: &mut Bencher) {
let s = "предложение из кириллических букв.";
assert_fmt(s, r#""предложение из кириллических букв.""#, 3);
b.iter(|| {
black_box(format!("{:?}", black_box(s)));
});
}

#[bench]
fn mixed(b: &mut Bencher) {
let s = "\"❤️\"\n\"hűha ez betű\"\n\"кириллических букв\".";
assert_fmt(s, r#""\"❤\u{fe0f}\"\n\"hűha ez betű\"\n\"кириллических букв\".""#, 36);
b.iter(|| {
black_box(format!("{:?}", black_box(s)));
});
}
Loading