Skip to content

ARROW-10810: [Rust] Speed up comparison kernels #8837

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
Dandandan wants to merge 3 commits into from
Closed

ARROW-10810: [Rust] Speed up comparison kernels #8837

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
32344ef
Comparison kernel speedup
Dandandan Dec 5, 2020
55f3820
Make it hae same sizing as the bufferbuilder
Dandandan Dec 5, 2020
62e71a2
Simplify
Dandandan Dec 5, 2020
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
30 changes: 24 additions & 6 deletions rust/arrow/src/compute/kernels/comparison.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,9 +47,18 @@ macro_rules! compare_op {
let null_bit_buffer =
combine_option_bitmap($left.data_ref(), $right.data_ref(), $left.len())?;

let mut result = BooleanBufferBuilder::new($left.len());
let byte_capacity = bit_util::ceil($left.len(), 8);
let actual_capacity = bit_util::round_upto_multiple_of_64(byte_capacity);
let mut buffer = MutableBuffer::new(actual_capacity);
buffer.resize(byte_capacity);

let data = buffer.raw_data_mut();
for i in 0..$left.len() {
result.append($op($left.value(i), $right.value(i)))?;
if $op($left.value(i), $right.value(i)) {
unsafe {
bit_util::set_bit_raw(data, i);
}
}
}

let data = ArrayData::new(
Expand All @@ -58,7 +67,7 @@ macro_rules! compare_op {
None,
null_bit_buffer,
0,
vec![result.finish()],
vec![buffer.freeze()],
vec![],
);
Ok(PrimitiveArray::<BooleanType>::from(Arc::new(data)))
Expand All @@ -68,9 +77,18 @@ macro_rules! compare_op {
macro_rules! compare_op_scalar {
($left: expr, $right:expr, $op:expr) => {{
let null_bit_buffer = $left.data().null_buffer().cloned();
let mut result = BooleanBufferBuilder::new($left.len());
let byte_capacity = bit_util::ceil($left.len(), 8);
let actual_capacity = bit_util::round_upto_multiple_of_64(byte_capacity);
let mut buffer = MutableBuffer::new(actual_capacity);
buffer.resize(byte_capacity);

let data = buffer.raw_data_mut();
for i in 0..$left.len() {
result.append($op($left.value(i), $right))?;
if $op($left.value(i), $right) {
unsafe {
bit_util::set_bit_raw(data, i);
}
}
Comment on lines -71 to +91
Copy link
Member

@jorgecarleitao jorgecarleitao Dec 5, 2020

Choose a reason for hiding this comment

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

A side note: In general, if we are bypassing builders for speed, it can mean that builders are not a good abstraction, as they significantly impact performance.

Copy link
Contributor Author

@Dandandan Dandandan Dec 5, 2020

Choose a reason for hiding this comment

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

Yes, would be nice to see if we can get the combination of a clean/safe API and good performance as well.

}

let data = ArrayData::new(
Expand All @@ -79,7 +97,7 @@ macro_rules! compare_op_scalar {
None,
null_bit_buffer,
0,
vec![result.finish()],
vec![buffer.freeze()],
vec![],
);
Ok(PrimitiveArray::<BooleanType>::from(Arc::new(data)))
Expand Down