This is an automated email from the ASF dual-hosted git repository.
alamb pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow-rs.git
The following commit(s) were added to refs/heads/main by this push:
new 1d857758a1 [10125] Introduce mult-batch decode benchmarks (#10207)
1d857758a1 is described below
commit 1d857758a14756e47be31e27b49ac5073e394120
Author: RIchard Baah <[email protected]>
AuthorDate: Wed Jun 24 18:53:16 2026 -0400
[10125] Introduce mult-batch decode benchmarks (#10207)
# Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and
enhancements and this helps us generate change logs for our releases.
You can link an issue to this PR using the GitHub syntax.
-->
Follow up of https://github.com/apache/arrow-rs/pull/10202 , see
[comment](https://github.com/apache/arrow-rs/pull/10202#discussion_r3462450393)
works towards #10125
# Rationale for this change
this provides more visibility to performances changes that occurs on the
decode path. Multi batch decoding is more representative of real world
work loads.
<!--
Why are you proposing this change? If this is already explained clearly
in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand
your changes and offer better suggestions for fixes.
-->
# What changes are included in this PR?
Added multi-batch decode benchmarks covering both plain and dictionary
types, including the replacement-dictionary path,
<!--
There is no need to duplicate the description in the issue here but it
is sometimes worth providing a summary of the individual changes in this
PR.
-->
# Are these changes tested?
n/a
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example,
are they covered by existing tests)?
If this PR claims a performance improvement, please include evidence
such as benchmark results.
-->
# Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be
updated before approving the PR.
If there are any breaking changes to public APIs, please call them out.
-->
no
---
arrow-flight/benches/flight.rs | 60 +++++++++++++++++++++++++++++++++++++++---
1 file changed, 57 insertions(+), 3 deletions(-)
diff --git a/arrow-flight/benches/flight.rs b/arrow-flight/benches/flight.rs
index 3519b8e133..1c6e50dad6 100644
--- a/arrow-flight/benches/flight.rs
+++ b/arrow-flight/benches/flight.rs
@@ -30,6 +30,7 @@ use common::{DICT_TYPES, TYPES, build_batch, start_server};
const ROWS: [usize; 2] = [8 * 1024, 64 * 1024];
const COLS: [usize; 3] = [1, 4, 8];
+const BATCHES: usize = 4;
fn bench_encode(c: &mut Criterion) {
let rt = tokio::runtime::Runtime::new().unwrap();
@@ -74,16 +75,22 @@ fn bench_decode(c: &mut Criterion) {
for &(name, build) in TYPES {
for &rows in &ROWS {
for &cols in &COLS {
- let batch = build_batch(name, rows, cols, build);
+ let batches: Vec<RecordBatch> = (0..BATCHES)
+ .map(|_| build_batch(name, rows, cols, build))
+ .collect();
+ let total_bytes: u64 = batches
+ .iter()
+ .map(|b| b.get_array_memory_size() as u64)
+ .sum();
let frames: Vec<FlightData> = rt
.block_on(
FlightDataEncoderBuilder::new()
- .build(futures::stream::iter([Ok(batch.clone())]))
+
.build(futures::stream::iter(batches.into_iter().map(Ok)))
.try_collect(),
)
.unwrap();
let id = BenchmarkId::new(name, format!("{rows}x{cols}"));
- g.throughput(Throughput::Bytes(batch.get_array_memory_size()
as u64));
+ g.throughput(Throughput::Bytes(total_bytes));
g.bench_function(id, |b| {
b.to_async(&rt).iter_batched(
|| frames.clone(),
@@ -124,6 +131,52 @@ fn bench_roundtrip(c: &mut Criterion) {
}
}
+/// Decode a multi-batch stream covering both plain and dictionary types.
+/// Uses [`DictionaryHandling::Resend`] to exercise the replacement-dictionary
path.
+fn bench_decode_dictionary_stream(c: &mut Criterion) {
+ let rt = tokio::runtime::Runtime::new().unwrap();
+ let mut g = c.benchmark_group("decode_stream");
+
+ for &(name, build) in TYPES.iter().chain(DICT_TYPES) {
+ for &rows in &ROWS {
+ for &cols in &COLS {
+ let batches: Vec<RecordBatch> = (0..BATCHES)
+ .map(|_| build_batch(name, rows, cols, build))
+ .collect();
+ let total_bytes: u64 = batches
+ .iter()
+ .map(|b| b.get_array_memory_size() as u64)
+ .sum();
+ let frames: Vec<FlightData> = rt
+ .block_on(
+ FlightDataEncoderBuilder::new()
+
.with_dictionary_handling(DictionaryHandling::Resend)
+
.build(futures::stream::iter(batches.into_iter().map(Ok)))
+ .try_collect(),
+ )
+ .unwrap();
+ let id = BenchmarkId::new(name,
format!("{rows}x{cols}x{BATCHES}"));
+ g.throughput(Throughput::Bytes(total_bytes));
+ g.bench_function(id, |b| {
+ b.to_async(&rt).iter_batched(
+ || frames.clone(),
+ |frames| async move {
+ let _: Vec<RecordBatch> =
+ FlightRecordBatchStream::new_from_flight_data(
+
futures::stream::iter(frames.into_iter().map(Ok)),
+ )
+ .try_collect()
+ .await
+ .unwrap();
+ },
+ criterion::BatchSize::SmallInput,
+ );
+ });
+ }
+ }
+ }
+}
+
fn bench_do_put_dictionary(c: &mut Criterion) {
let rt = tokio::runtime::Runtime::new().unwrap();
let (channel, _) = rt.block_on(start_server());
@@ -173,6 +226,7 @@ criterion_group!(
benches,
bench_encode,
bench_decode,
+ bench_decode_dictionary_stream,
bench_roundtrip,
bench_do_put_dictionary
);
