Skip to content

ARROW-11630: [Rust] Introduce limit option for sort kernel #9602

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.

Sign up for GitHub

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

Closed
sundy-li wants to merge 11 commits into from
Closed

ARROW-11630: [Rust] Introduce limit option for sort kernel #9602

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
abfb7c9
[Rust] Introduce limit option for sort kernel
sundy-li Feb 28, 2021
06c503c
Add tests and benchmarks
sundy-li Mar 1, 2021
c7df0d6
update comment
sundy-li Mar 1, 2021
4f4dc62
update styles
sundy-li Mar 1, 2021
dc3167e
Remove pdqsort && add a separate function partial_sort
sundy-li Mar 2, 2021
b98b0b3
1. Introduce a wrap function for sort and partial_sort, if limit size
sundy-li Mar 4, 2021
bb6a77e
Update tests
sundy-li Mar 7, 2021
0461668
Include partial_sort to arrow
sundy-li Mar 8, 2021
86672cc
Make clippy test happy
sundy-li Mar 8, 2021
2dd16a7
partial_sort from select_nth_unstable_by + sort
sundy-li Mar 12, 2021
fd6a234
Add docs about unstable_sort
sundy-li Mar 14, 2021
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
59 changes: 53 additions & 6 deletions rust/arrow/benches/sort_kernel.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,7 +33,7 @@ fn create_array(size: usize, with_nulls: bool) -> ArrayRef {
Arc::new(array)
}

fn bench_sort(arr_a: &ArrayRef, array_b: &ArrayRef) {
fn bench_sort(arr_a: &ArrayRef, array_b: &ArrayRef, limit: Option<usize>) {
let columns = vec![
SortColumn {
values: arr_a.clone(),
Expand All @@ -45,29 +45,76 @@ fn bench_sort(arr_a: &ArrayRef, array_b: &ArrayRef) {
},
];

criterion::black_box(lexsort(&columns).unwrap());
criterion::black_box(lexsort(&columns, limit).unwrap());
}

fn add_benchmark(c: &mut Criterion) {
let arr_a = create_array(2u64.pow(10) as usize, false);
let arr_b = create_array(2u64.pow(10) as usize, false);

c.bench_function("sort 2^10", |b| b.iter(|| bench_sort(&arr_a, &arr_b)));
c.bench_function("sort 2^10", |b| b.iter(|| bench_sort(&arr_a, &arr_b, None)));

let arr_a = create_array(2u64.pow(12) as usize, false);
let arr_b = create_array(2u64.pow(12) as usize, false);

c.bench_function("sort 2^12", |b| b.iter(|| bench_sort(&arr_a, &arr_b)));
c.bench_function("sort 2^12", |b| b.iter(|| bench_sort(&arr_a, &arr_b, None)));

let arr_a = create_array(2u64.pow(10) as usize, true);
let arr_b = create_array(2u64.pow(10) as usize, true);

c.bench_function("sort nulls 2^10", |b| b.iter(|| bench_sort(&arr_a, &arr_b)));
c.bench_function("sort nulls 2^10", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, None))
});

let arr_a = create_array(2u64.pow(12) as usize, true);
let arr_b = create_array(2u64.pow(12) as usize, true);

c.bench_function("sort nulls 2^12", |b| b.iter(|| bench_sort(&arr_a, &arr_b)));
c.bench_function("sort nulls 2^12", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, None))
});

// with limit
{
let arr_a = create_array(2u64.pow(12) as usize, false);
let arr_b = create_array(2u64.pow(12) as usize, false);
c.bench_function("sort 2^12 limit 10", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(10)))
});

let arr_a = create_array(2u64.pow(12) as usize, false);
let arr_b = create_array(2u64.pow(12) as usize, false);
c.bench_function("sort 2^12 limit 100", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(100)))
});

let arr_a = create_array(2u64.pow(12) as usize, false);
let arr_b = create_array(2u64.pow(12) as usize, false);
c.bench_function("sort 2^12 limit 1000", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(1000)))
});

let arr_a = create_array(2u64.pow(12) as usize, false);
let arr_b = create_array(2u64.pow(12) as usize, false);
c.bench_function("sort 2^12 limit 2^12", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(2u64.pow(12) as usize)))
});

let arr_a = create_array(2u64.pow(12) as usize, true);
let arr_b = create_array(2u64.pow(12) as usize, true);

c.bench_function("sort nulls 2^12 limit 10", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(10)))
});
c.bench_function("sort nulls 2^12 limit 100", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(100)))
});
c.bench_function("sort nulls 2^12 limit 1000", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(1000)))
});
c.bench_function("sort nulls 2^12 limit 2^12", |b| {
b.iter(|| bench_sort(&arr_a, &arr_b, Some(2u64.pow(12) as usize)))
});
}
}

criterion_group!(benches, add_benchmark);
Expand Down
Loading