Improve performance of `rem_euclid()` for signed integers #103913

Neutron3529 · 2022-11-03T08:35:54Z

such code is copy from
https://github.com/rust-lang/rust/blob/master/library/std/src/f32.rs and
https://github.com/rust-lang/rust/blob/master/library/std/src/f64.rs
using r+rhs.abs() is faster than calc it with an if clause. Bench result:

$ cargo bench
   Compiling div-euclid v0.1.0 (/me/div-euclid)
    Finished bench [optimized] target(s) in 1.01s
     Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef)

running 7 tests
test tests::it_works ... ignored
test tests::bench_aaabs     ... bench:  10,498,793 ns/iter (+/- 104,360)
test tests::bench_aadefault ... bench:  11,061,862 ns/iter (+/- 94,107)
test tests::bench_abs       ... bench:  10,477,193 ns/iter (+/- 81,942)
test tests::bench_default   ... bench:  10,622,983 ns/iter (+/- 25,119)
test tests::bench_zzabs     ... bench:  10,481,971 ns/iter (+/- 43,787)
test tests::bench_zzdefault ... bench:  11,074,976 ns/iter (+/- 29,633)

test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s

It seems that, default rem_euclid triggered a branch prediction, thus bench_default is faster than bench_aadefault and bench_aadefault, which shuffles the order of calculations. but all of them slower than what it was in f64's and f32's rem_euclid, thus I submit this PR.

bench code:

#![feature(test)]
extern crate test;

fn rem_euclid(a:i32,rhs:i32)->i32{
    let r = a % rhs;
    if r < 0 { r + rhs.abs() } else { r }
}

#[cfg(test)]
mod tests {
    use super::*;
    use test::Bencher;
    use rand::prelude::*;
    use rand::rngs::SmallRng;
    const N:i32=1000;
    #[test]
    fn it_works() {
        let a: i32 = 7; // or any other integer type
        let b = 4;

        let d:Vec<i32>=(-N..=N).collect();
        let n:Vec<i32>=(-N..0).chain(1..=N).collect();

        for i in &d {
            for j in &n {
                assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j));
            }
        }

        assert_eq!(rem_euclid(a,b), 3);
        assert_eq!(rem_euclid(-a,b), 1);
        assert_eq!(rem_euclid(a,-b), 3);
        assert_eq!(rem_euclid(-a,-b), 1);
    }


    #[bench]
    fn bench_aaabs(b: &mut Bencher) {
        let mut d:Vec<i32>=(-N..=N).collect();
        let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
        let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
        n.shuffle(&mut rng);
        d.shuffle(&mut rng);
        n.shuffle(&mut rng);
        b.iter(||{
            let mut res=0;
            for i in &d {
                for j in &n {
                    res+=rem_euclid(*i,*j);
                }
            }
            res
        });
    }
    #[bench]
    fn bench_aadefault(b: &mut Bencher) {
        let mut d:Vec<i32>=(-N..=N).collect();
        let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
        let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
        n.shuffle(&mut rng);
        d.shuffle(&mut rng);
        n.shuffle(&mut rng);
        b.iter(||{
            let mut res=0;
            for i in &d {
                for j in &n {
                    res+=i.rem_euclid(*j);
                }
            }
            res
        });
    }

    #[bench]
    fn bench_abs(b: &mut Bencher) {
        let d:Vec<i32>=(-N..=N).collect();
        let n:Vec<i32>=(-N..0).chain(1..=N).collect();
        b.iter(||{
            let mut res=0;
            for i in &d {
                for j in &n {
                    res+=rem_euclid(*i,*j);
                }
            }
            res
        });
    }
    #[bench]
    fn bench_default(b: &mut Bencher) {
        let d:Vec<i32>=(-N..=N).collect();
        let n:Vec<i32>=(-N..0).chain(1..=N).collect();
        b.iter(||{
            let mut res=0;
            for i in &d {
                for j in &n {
                    res+=i.rem_euclid(*j);
                }
            }
            res
        });
    }

    #[bench]
    fn bench_zzabs(b: &mut Bencher) {
        let mut d:Vec<i32>=(-N..=N).collect();
        let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
        let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
        d.shuffle(&mut rng);
        n.shuffle(&mut rng);
        d.shuffle(&mut rng);
        b.iter(||{
            let mut res=0;
            for i in &d {
                for j in &n {
                    res+=rem_euclid(*i,*j);
                }
            }
            res
        });
    }
    #[bench]
    fn bench_zzdefault(b: &mut Bencher) {
        let mut d:Vec<i32>=(-N..=N).collect();
        let mut n:Vec<i32>=(-N..0).chain(1..=N).collect();
        let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]);
        d.shuffle(&mut rng);
        n.shuffle(&mut rng);
        d.shuffle(&mut rng);
        b.iter(||{
            let mut res=0;
            for i in &d {
                for j in &n {
                    res+=i.rem_euclid(*j);
                }
            }
            res
        });
    }
}

such code is copy from https://github.com/rust-lang/rust/blob/master/library/std/src/f32.rs and https://github.com/rust-lang/rust/blob/master/library/std/src/f64.rs using r+rhs.abs() is faster than calc it directly. Bench result: ``` $ cargo bench Compiling div-euclid v0.1.0 (/me/div-euclid) Finished bench [optimized] target(s) in 1.01s Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef) running 7 tests test tests::it_works ... ignored test tests::bench_aaabs ... bench: 10,498,793 ns/iter (+/- 104,360) test tests::bench_aadefault ... bench: 11,061,862 ns/iter (+/- 94,107) test tests::bench_abs ... bench: 10,477,193 ns/iter (+/- 81,942) test tests::bench_default ... bench: 10,622,983 ns/iter (+/- 25,119) test tests::bench_zzabs ... bench: 10,481,971 ns/iter (+/- 43,787) test tests::bench_zzdefault ... bench: 11,074,976 ns/iter (+/- 29,633) test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s ``` bench code: ``` #![feature(test)] extern crate test; fn rem_euclid(a:i32,rhs:i32)->i32{ let r = a % rhs; if r < 0 { r + rhs.abs() } else { r } } #[cfg(test)] mod tests { use super::*; use test::Bencher; use rand::prelude::*; use rand::rngs::SmallRng; const N:i32=1000; #[test] fn it_works() { let a: i32 = 7; // or any other integer type let b = 4; let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); for i in &d { for j in &n { assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j)); } } assert_eq!(rem_euclid(a,b), 3); assert_eq!(rem_euclid(-a,b), 1); assert_eq!(rem_euclid(a,-b), 3); assert_eq!(rem_euclid(-a,-b), 1); } #[bench] fn bench_aaabs(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); n.shuffle(&mut rng); d.shuffle(&mut rng); n.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_aadefault(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); n.shuffle(&mut rng); d.shuffle(&mut rng); n.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } #[bench] fn bench_abs(b: &mut Bencher) { let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_default(b: &mut Bencher) { let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } #[bench] fn bench_zzabs(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); d.shuffle(&mut rng); n.shuffle(&mut rng); d.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_zzdefault(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); d.shuffle(&mut rng); n.shuffle(&mut rng); d.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } } ```

rustbot · 2022-11-03T08:36:01Z

r? @thomcc

(rustbot has picked a reviewer for you, use r? to override)

rustbot · 2022-11-03T08:36:04Z

Hey! It looks like you've submitted a new PR for the library teams!

If this PR contains changes to any rust-lang/rust public library APIs then please comment with @rustbot label +T-libs-api -T-libs to tag it appropriately. If this PR contains changes to any unstable APIs please edit the PR description to add a link to the relevant API Change Proposal or create one if you haven't already. If you're unsure where your change falls no worries, just leave it as is and the reviewer will take a look and make a decision to forward on if necessary.

Examples of T-libs-api changes:

Stabilizing library features
Introducing insta-stable changes such as new implementations of existing stable traits on existing stable types
Introducing new or changing existing unstable library APIs (excluding permanently unstable features / features without a tracking issue)
Changing public documentation in ways that create new stability guarantees
Changing observable runtime behavior of library APIs

benchmark result: ``` $ cargo bench Compiling div-euclid v0.1.0 (/me/div-euclid) Finished bench [optimized] target(s) in 1.01s Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef) running 7 tests test tests::it_works ... ignored test tests::bench_aaabs ... bench: 10,498,793 ns/iter (+/- 104,360) test tests::bench_aadefault ... bench: 11,061,862 ns/iter (+/- 94,107) test tests::bench_abs ... bench: 10,477,193 ns/iter (+/- 81,942) test tests::bench_default ... bench: 10,622,983 ns/iter (+/- 25,119) test tests::bench_zzabs ... bench: 10,481,971 ns/iter (+/- 43,787) test tests::bench_zzdefault ... bench: 11,074,976 ns/iter (+/- 29,633) test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s ``` benchmark code: ```rust #![feature(test)] extern crate test; #[inline(always)] fn rem_euclid(a:i32,rhs:i32)->i32{ let r = a % rhs; if r < 0 { // if rhs is `integer::MIN`, rhs.wrapping_abs() == rhs.wrapping_abs, // thus r.wrapping_add(rhs.wrapping_abs()) == r.wrapping_add(rhs) == r - rhs, // which suits our need. // otherwise, rhs.wrapping_abs() == -rhs, which won't overflow since r is negative. r.wrapping_add(rhs.wrapping_abs()) } else { r } } #[cfg(test)] mod tests { use super::*; use test::Bencher; use rand::prelude::*; use rand::rngs::SmallRng; const N:i32=1000; #[test] fn it_works() { let a: i32 = 7; // or any other integer type let b = 4; let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); for i in &d { for j in &n { assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j)); } } assert_eq!(rem_euclid(a,b), 3); assert_eq!(rem_euclid(-a,b), 1); assert_eq!(rem_euclid(a,-b), 3); assert_eq!(rem_euclid(-a,-b), 1); } #[bench] fn bench_aaabs(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); n.shuffle(&mut rng); d.shuffle(&mut rng); n.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_aadefault(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); n.shuffle(&mut rng); d.shuffle(&mut rng); n.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } #[bench] fn bench_abs(b: &mut Bencher) { let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_default(b: &mut Bencher) { let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } #[bench] fn bench_zzabs(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); d.shuffle(&mut rng); n.shuffle(&mut rng); d.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_zzdefault(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); d.shuffle(&mut rng); n.shuffle(&mut rng); d.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } } ```

thomcc

This looks good and is definitely a correct change, but i think the comment needs work. I gave a suggestion which you can take or not.

library/core/src/num/int_macros.rs

thomcc · 2022-11-04T04:26:14Z

I was not able to reproduce these benchmark numbers locally. I actually found your version consistently slightly slower. Note that wrapping_abs contains a branch.

In your benchmark, the outer loop has the same value every iteration -- when checking the version that is hard on the branch predictor, you should be sure to shuffle the groups of (a, b), for example:

fn samples() -> Vec<(i32, i32)> {
	let n = 8000;
    let mut all = (-n..=n)
        .flat_map(|n| (-n..=n).map(move |d| (n, d)))
        .filter(|(n, d)| n.checked_rem(*d).is_some())
        .collect::<Vec<_>>();
    let mut r = SmallRng::from_seed(SEED);
    all.shuffle(&mut r);
    all
}

@rustbot author

thomcc · 2022-11-04T04:36:48Z

Actually, this is a more decisive win on x86, although it's still slight. Hmm.

thomcc · 2022-11-04T04:44:23Z

My benchmark seems around ~5% slower on aarch64, but has smaller code size. On x86_64 it's 10% faster with much cleaner/smaller assembly and one fewer potentially-unpredictable branch. I think we should probably take it.

Please update the comment, though.

I'll trigger a perf run too, just in case this happens to be used in some weird data structure in the compiler.

@bors try @rust-timer queue

rust-timer · 2022-11-04T04:44:26Z

Awaiting bors try build completion.

@rustbot label: +S-waiting-on-perf

bors · 2022-11-04T04:44:33Z

⌛ Trying commit aafe6db with merge 4e489a0c24dfe15e921c177ef0573acc977c8954...

bors · 2022-11-04T06:48:54Z

☀️ Try build successful - checks-actions
Build commit: 4e489a0c24dfe15e921c177ef0573acc977c8954 (4e489a0c24dfe15e921c177ef0573acc977c8954)

rust-timer · 2022-11-04T06:48:56Z

Queued 4e489a0c24dfe15e921c177ef0573acc977c8954 with parent 6718ea1, future comparison URL.

Neutron3529 · 2022-11-04T08:01:11Z

Please update the comment, though.

The comment is updated.

rust-timer · 2022-11-04T08:07:04Z

Finished benchmarking commit (4e489a0c24dfe15e921c177ef0573acc977c8954): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

@bors rollup=never
@rustbot label: +S-waiting-on-review -S-waiting-on-perf -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

This benchmark run did not return any relevant results for this metric.

Cycles

This benchmark run did not return any relevant results for this metric.

thomcc · 2022-11-12T02:04:54Z

@bors r+

bors · 2022-11-12T02:04:56Z

📌 Commit d81a0e9 has been approved by thomcc

It is now in the queue for this repository.

bors · 2022-11-12T20:48:30Z

⌛ Testing commit d81a0e9 with merge 6284998...

bors · 2022-11-12T23:29:05Z

☀️ Test successful - checks-actions
Approved by: thomcc
Pushing 6284998 to master...

rust-timer · 2022-11-13T00:47:15Z

Finished benchmarking commit (6284998): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.8%	[3.8%, 3.8%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-6.3%	[-6.3%, -6.3%]	1
All ❌✅ (primary)	-	-	0

Cycles

This benchmark run did not return any relevant results for this metric.

Improve performance of `rem_euclid()` for signed integers such code is copy from https://github.com/rust-lang/rust/blob/master/library/std/src/f32.rs and https://github.com/rust-lang/rust/blob/master/library/std/src/f64.rs using `r+rhs.abs()` is faster than calc it with an if clause. Bench result: ``` $ cargo bench Compiling div-euclid v0.1.0 (/me/div-euclid) Finished bench [optimized] target(s) in 1.01s Running unittests src/lib.rs (target/release/deps/div_euclid-7a4530ca7817d1ef) running 7 tests test tests::it_works ... ignored test tests::bench_aaabs ... bench: 10,498,793 ns/iter (+/- 104,360) test tests::bench_aadefault ... bench: 11,061,862 ns/iter (+/- 94,107) test tests::bench_abs ... bench: 10,477,193 ns/iter (+/- 81,942) test tests::bench_default ... bench: 10,622,983 ns/iter (+/- 25,119) test tests::bench_zzabs ... bench: 10,481,971 ns/iter (+/- 43,787) test tests::bench_zzdefault ... bench: 11,074,976 ns/iter (+/- 29,633) test result: ok. 0 passed; 0 failed; 1 ignored; 6 measured; 0 filtered out; finished in 19.35s ``` It seems that, default `rem_euclid` triggered a branch prediction, thus `bench_default` is faster than `bench_aadefault` and `bench_aadefault`, which shuffles the order of calculations. but all of them slower than what it was in `f64`'s and `f32`'s `rem_euclid`, thus I submit this PR. bench code: ```rust #![feature(test)] extern crate test; fn rem_euclid(a:i32,rhs:i32)->i32{ let r = a % rhs; if r < 0 { r + rhs.abs() } else { r } } #[cfg(test)] mod tests { use super::*; use test::Bencher; use rand::prelude::*; use rand::rngs::SmallRng; const N:i32=1000; #[test] fn it_works() { let a: i32 = 7; // or any other integer type let b = 4; let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); for i in &d { for j in &n { assert_eq!(i.rem_euclid(*j),rem_euclid(*i,*j)); } } assert_eq!(rem_euclid(a,b), 3); assert_eq!(rem_euclid(-a,b), 1); assert_eq!(rem_euclid(a,-b), 3); assert_eq!(rem_euclid(-a,-b), 1); } #[bench] fn bench_aaabs(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); n.shuffle(&mut rng); d.shuffle(&mut rng); n.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_aadefault(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); n.shuffle(&mut rng); d.shuffle(&mut rng); n.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } #[bench] fn bench_abs(b: &mut Bencher) { let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_default(b: &mut Bencher) { let d:Vec<i32>=(-N..=N).collect(); let n:Vec<i32>=(-N..0).chain(1..=N).collect(); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } #[bench] fn bench_zzabs(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); d.shuffle(&mut rng); n.shuffle(&mut rng); d.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=rem_euclid(*i,*j); } } res }); } #[bench] fn bench_zzdefault(b: &mut Bencher) { let mut d:Vec<i32>=(-N..=N).collect(); let mut n:Vec<i32>=(-N..0).chain(1..=N).collect(); let mut rng=SmallRng::from_seed([1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,21]); d.shuffle(&mut rng); n.shuffle(&mut rng); d.shuffle(&mut rng); b.iter(||{ let mut res=0; for i in &d { for j in &n { res+=i.rem_euclid(*j); } } res }); } } ```

rustbot assigned thomcc Nov 3, 2022

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Nov 3, 2022

This comment has been minimized.

Sign in to view

Neutron3529 changed the title ~~Optimize the code to run faster.~~ faster div-euclid for int Nov 4, 2022

thomcc reviewed Nov 4, 2022

View reviewed changes

library/core/src/num/int_macros.rs Outdated Show resolved Hide resolved

thomcc changed the title ~~faster div-euclid for int~~ Improve performance of rem_euclid() for signed integers Nov 4, 2022

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Nov 4, 2022

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Nov 4, 2022

update comment

d81a0e9

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Nov 4, 2022

bors removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Nov 12, 2022

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Nov 12, 2022

bors mentioned this pull request Nov 12, 2022

linker: Refactoring and fixes to native library linking #103311

Merged

bors added the merged-by-bors This PR was explicitly merged by bors. label Nov 12, 2022

bors merged commit 6284998 into rust-lang:master Nov 12, 2022

rustbot added this to the 1.67.0 milestone Nov 12, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Improve performance of `rem_euclid()` for signed integers #103913

Improve performance of `rem_euclid()` for signed integers #103913

Neutron3529 commented Nov 3, 2022 •

edited

Loading

rustbot commented Nov 3, 2022

rustbot commented Nov 3, 2022

This comment has been minimized.

thomcc left a comment

thomcc commented Nov 4, 2022

thomcc commented Nov 4, 2022

thomcc commented Nov 4, 2022

rust-timer commented Nov 4, 2022

bors commented Nov 4, 2022

bors commented Nov 4, 2022

rust-timer commented Nov 4, 2022

Neutron3529 commented Nov 4, 2022

rust-timer commented Nov 4, 2022

thomcc commented Nov 12, 2022

bors commented Nov 12, 2022

bors commented Nov 12, 2022

bors commented Nov 12, 2022

rust-timer commented Nov 13, 2022

Improve performance of rem_euclid() for signed integers #103913

Improve performance of rem_euclid() for signed integers #103913

Conversation

Neutron3529 commented Nov 3, 2022 • edited Loading

rustbot commented Nov 3, 2022

rustbot commented Nov 3, 2022

This comment has been minimized.

thomcc left a comment

Choose a reason for hiding this comment

thomcc commented Nov 4, 2022

thomcc commented Nov 4, 2022

thomcc commented Nov 4, 2022

rust-timer commented Nov 4, 2022

bors commented Nov 4, 2022

bors commented Nov 4, 2022

rust-timer commented Nov 4, 2022

Neutron3529 commented Nov 4, 2022

rust-timer commented Nov 4, 2022

Overall result: no relevant changes - no action needed

Instruction count

Max RSS (memory usage)

Cycles

thomcc commented Nov 12, 2022

bors commented Nov 12, 2022

bors commented Nov 12, 2022

bors commented Nov 12, 2022

rust-timer commented Nov 13, 2022

Overall result: no relevant changes - no action needed

Instruction count

Max RSS (memory usage)

Cycles

Improve performance of `rem_euclid()` for signed integers #103913

Improve performance of `rem_euclid()` for signed integers #103913

Neutron3529 commented Nov 3, 2022 •

edited

Loading