Skip to content

bench: run morsel reader in CI benchmarks #9684

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

Draft
joseph-isaacs wants to merge 5 commits into
base: develop
Choose a base branch
from
Draft

bench: run morsel reader in CI benchmarks #9684

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
5b3cb2e
bench: exercise the morsel reader in CI
joseph-isaacs Aug 28, 2026
5ceeb22
bench: drive SQL scans with morsel executor
joseph-isaacs Aug 28, 2026
a1f479f
chore: fix prototype spelling checks
joseph-isaacs Aug 28, 2026
6131b16
perf: avoid redundant single-thread morsel pools
joseph-isaacs Aug 28, 2026
681bb3d
perf: share morsel scans across SQL batches
joseph-isaacs Aug 29, 2026
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
34 changes: 34 additions & 0 deletions Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

2 changes: 2 additions & 0 deletions Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@ members = [
"vortex-compressor",
"vortex-btrblocks",
"vortex-layout",
"vortex-morsel",
"vortex-scan",
"vortex-file",
"vortex-ipc",
Expand Down Expand Up @@ -325,6 +326,7 @@ vortex-pco = { version = "0.1.0", path = "./encodings/pco", default-features = f
vortex-proto = { version = "0.1.0", path = "./vortex-proto", default-features = false }
vortex-row = { version = "0.1.0", path = "./vortex-row", default-features = false }
vortex-runend = { version = "0.1.0", path = "./encodings/runend", default-features = false }
vortex-morsel = { version = "0.1.0", path = "./vortex-morsel", default-features = false }
vortex-scan = { version = "0.1.0", path = "./vortex-scan", default-features = false }
vortex-sequence = { version = "0.1.0", path = "encodings/sequence", default-features = false }
vortex-session = { version = "0.1.0", path = "./vortex-session", default-features = false }
Expand Down
1 change: 1 addition & 0 deletions benchmarks/compress-bench/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,6 +37,7 @@ vortex = { workspace = true }
vortex-arrow = { workspace = true }
vortex-bench = { workspace = true }
vortex-cuda = { workspace = true, optional = true }
vortex-morsel = { workspace = true }

[features]
cuda = ["dep:tempfile", "dep:vortex-cuda"]
Expand Down
142 changes: 82 additions & 60 deletions benchmarks/compress-bench/src/main.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -444,67 +444,89 @@ async fn run_benchmark_for_dataset(
let compressor = get_compressor(*format, mode);

for op in ops {
let time = match op {
CompressOp::Compress => {
let result = benchmark_compress(
compressor.as_ref(),
&parquet_path,
iterations,
bench_name,
)
.await
.with_context(|| format!("compressing {bench_name} as {format}"))?;
compressed_sizes.insert(*format, result.compressed_size);
let all_runs_ns: Vec<u64> = result
.all_runs
.iter()
.map(|d| u64::try_from(d.as_nanos()).unwrap_or(u64::MAX))
.collect();
v3_records.push(v3::compression_time_record(
&result.timing,
v3_dataset,
v3_variant,
CompressOp::Compress,
all_runs_ns,
));
v3_records.push(v3::compression_size_record(
v3_dataset,
v3_variant,
*format,
result.compressed_size,
uncompressed_size.context("compression size requires Arrow memory size")?,
));
ratios.extend(result.ratios);
timings.push(result.timing);
result.time
let run = AssertUnwindSafe(async {
anyhow::Ok(match op {
CompressOp::Compress => {
let result = benchmark_compress(
compressor.as_ref(),
&parquet_path,
iterations,
bench_name,
)
.await
.with_context(|| format!("compressing {bench_name} as {format}"))?;
compressed_sizes.insert(*format, result.compressed_size);
let all_runs_ns: Vec<u64> = result
.all_runs
.iter()
.map(|d| u64::try_from(d.as_nanos()).unwrap_or(u64::MAX))
.collect();
v3_records.push(v3::compression_time_record(
&result.timing,
v3_dataset,
v3_variant,
CompressOp::Compress,
all_runs_ns,
));
v3_records.push(v3::compression_size_record(
v3_dataset,
v3_variant,
*format,
result.compressed_size,
uncompressed_size
.context("compression size requires Arrow memory size")?,
));
ratios.extend(result.ratios);
timings.push(result.timing);
result.time
}
CompressOp::Decompress => {
let result = benchmark_decompress(
compressor.as_ref(),
&parquet_path,
iterations,
&decompress_name,
)
.await
.with_context(|| format!("decompressing {bench_name} as {format}"))?;
let all_runs_ns: Vec<u64> = result
.all_runs
.iter()
.map(|d| u64::try_from(d.as_nanos()).unwrap_or(u64::MAX))
.collect();
v3_records.push(v3::compression_time_record(
&result.timing,
v3_dataset,
if mode.is_gpu() {
Some("gpu")
} else {
v3_variant
},
CompressOp::Decompress,
all_runs_ns,
));
timings.push(result.timing);
result.time
}
})
})
.catch_unwind()
.await;

let time = match run {
Ok(Ok(time)) => time,
Ok(Err(error)) => {
tracing::error!("dropping {op} result for {bench_name} as {format}: {error:#}");
progress.inc(1);
continue;
}
CompressOp::Decompress => {
let result = benchmark_decompress(
compressor.as_ref(),
&parquet_path,
iterations,
&decompress_name,
)
.await
.with_context(|| format!("decompressing {bench_name} as {format}"))?;
let all_runs_ns: Vec<u64> = result
.all_runs
.iter()
.map(|d| u64::try_from(d.as_nanos()).unwrap_or(u64::MAX))
.collect();
v3_records.push(v3::compression_time_record(
&result.timing,
v3_dataset,
if mode.is_gpu() {
Some("gpu")
} else {
v3_variant
},
CompressOp::Decompress,
all_runs_ns,
));
timings.push(result.timing);
result.time
Err(panic) => {
tracing::error!(
"dropping {op} result for {bench_name} as {format}: panicked: {}",
panic_message(&panic)
);
progress.inc(1);
continue;
}
};

Expand Down
48 changes: 31 additions & 17 deletions benchmarks/compress-bench/src/vortex.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,20 +10,26 @@ use std::time::Instant;
use anyhow::Result;
use async_trait::async_trait;
use bytes::Bytes;
use futures::StreamExt;
use futures::pin_mut;
use vortex::array::Canonical;
use vortex::array::IntoArray;
use vortex::array::VortexSessionExecute;
use vortex::dtype::FieldNames;
use vortex::expr::root;
use vortex::expr::select;
use vortex::file::OpenOptionsSessionExt;
use vortex::file::WriteOptionsSessionExt;
use vortex_arrow::ArrowSessionExt;
use vortex::utils::parallelism::get_available_parallelism;
use vortex_bench::Format;
use vortex_bench::SESSION;
use vortex_bench::compress::Compressor;
use vortex_bench::compress::read_projection;
use vortex_bench::conversions::parquet_to_vortex_chunks;
use vortex_morsel::MorselScan;
use vortex_morsel::build_plan;
use vortex_morsel::morsels;
use vortex_morsel::nodes::ConjunctMode;

const MORSEL_ROWS: u64 = 131_072;

/// Compressor implementation for Vortex format.
pub struct VortexCompressor;
Expand Down Expand Up @@ -63,26 +69,34 @@ impl Compressor for VortexCompressor {
// Now decompress
let start = Instant::now();
let data = Bytes::from(buf);
let mut scan = SESSION.open_options().open_buffer(data)?.scan()?;
let source_dtype = scan.dtype()?;
let file = SESSION.open_options().open_buffer(data)?;
let source_dtype = file.dtype().clone();
let root_columns = source_dtype
.as_struct_fields_opt()
.map_or(0, |fields| fields.nfields());
if let Some(cols) = read_projection(root_columns) {
let projection = if let Some(cols) = read_projection(root_columns) {
// Columns are named "0".."num_columns-1"; project the given subset.
let names: FieldNames = cols.iter().map(|i| i.to_string()).collect();
let projection = select(names, root())
.optimize_recursive(&source_dtype)?
.bind(&source_dtype)?;
scan = scan.with_projection(projection);
}
let schema = Arc::new(SESSION.arrow().to_arrow_schema(&scan.dtype()?)?);

let stream = scan.into_record_batch_stream(schema)?;
pin_mut!(stream);
select(names, root())
} else {
root()
};
let plan = Arc::new(build_plan(
file.footer().layout(),
&projection,
None,
ConjunctMode::Cascade,
)?);
let cut = morsels(&plan, MORSEL_ROWS);
let threads = get_available_parallelism().unwrap_or(1);
let (batches, _) = MorselScan::new(plan, file.segment_source(), SESSION.clone())
.with_threads(threads)
.with_morsels(cut)
.run()?;

while let Some(batch) = stream.next().await {
let _batch = batch?;
let mut ctx = SESSION.create_execution_ctx();
for batch in batches {
let _canonical = batch.execute::<Canonical>(&mut ctx)?;
}
Ok(start.elapsed())
}
Expand Down
1 change: 1 addition & 0 deletions docs/developer-guide/index.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ internals/async-runtime
internals/vtables
internals/execution
internals/scan-planning
internals/scan-execution-models/index
internals/stats-pruning
internals/io
internals/serialization
Expand Down
Loading
Loading