Skip to content

perf(encoding): measure compressed mini-block offsets #8333

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
Xuanwo wants to merge 1 commit into from
Draft

perf(encoding): measure compressed mini-block offsets #8333

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
61 changes: 44 additions & 17 deletions rust/lance-encoding/benches/common/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,11 +9,12 @@ use lance_encoding::{
compression::{
BlockCompressor, CompressionStrategy, field_metadata_params, finalize_miniblock_compressor,
reject_packed_struct_per_value, try_bitpacking_block, try_bitpacking_miniblock,
try_byte_stream_split_miniblock, try_fixed_packed_struct_miniblock, try_fixed_u8_rle_block,
try_fixed_u8_rle_miniblock, try_general_block, try_raw_block,
try_raw_fixed_size_list_miniblock, try_raw_fixed_width_miniblock, try_raw_per_value,
try_uncompressed_fixed_width_miniblock, try_variable_packed_struct_per_value,
try_variable_width_miniblock, try_variable_width_per_value,
try_byte_stream_split_miniblock, try_child_rle_miniblock,
try_fixed_packed_struct_miniblock, try_fixed_u8_rle_block, try_fixed_u8_rle_miniblock,
try_general_block, try_raw_block, try_raw_fixed_size_list_miniblock,
try_raw_fixed_width_miniblock, try_raw_per_value, try_uncompressed_fixed_width_miniblock,
try_variable_packed_struct_per_value, try_variable_rle_block, try_variable_width_miniblock,
try_variable_width_miniblock_with_generic_offsets, try_variable_width_per_value,
},
compression_config::{CompressionFieldParams, CompressionParams},
data::DataBlock,
Expand All @@ -33,6 +34,7 @@ pub enum BenchEncoding {
Array,
StructuralU16,
StructuralU32,
StructuralSparse,
}

impl std::fmt::Display for BenchEncoding {
Expand All @@ -41,6 +43,7 @@ impl std::fmt::Display for BenchEncoding {
Self::Array => "array",
Self::StructuralU16 => "structural-u16",
Self::StructuralU32 => "structural-u32",
Self::StructuralSparse => "structural-sparse",
})
}
}
Expand Down Expand Up @@ -81,13 +84,22 @@ impl CompressionStrategy for BenchCompressionStrategy {
compressor
} else if let Some(compressor) = try_byte_stream_split_miniblock(data, &params) {
compressor
} else if let Some(compressor) = try_fixed_u8_rle_miniblock(data, &params) {
} else if let Some(compressor) = match self.encoding {
BenchEncoding::StructuralSparse => try_child_rle_miniblock(data, &params),
BenchEncoding::Array
| BenchEncoding::StructuralU16
| BenchEncoding::StructuralU32 => try_fixed_u8_rle_miniblock(data, &params),
} {
compressor
} else if let Some(compressor) = try_bitpacking_miniblock(data) {
compressor
} else if let Some(compressor) = try_raw_fixed_width_miniblock(data) {
compressor
} else if let Some(compressor) = try_variable_width_miniblock(field, data, &params)? {
} else if let Some(compressor) = if self.encoding == BenchEncoding::StructuralSparse {
try_variable_width_miniblock_with_generic_offsets(field, data, &params)?
} else {
try_variable_width_miniblock(field, data, &params)?
} {
compressor
} else if let Some(compressor) = try_fixed_packed_struct_miniblock(data)? {
compressor
Expand Down Expand Up @@ -116,7 +128,7 @@ impl CompressionStrategy for BenchCompressionStrategy {
}
let packed = match self.encoding {
BenchEncoding::StructuralU16 => reject_packed_struct_per_value(field, data)?,
BenchEncoding::StructuralU32 => {
BenchEncoding::StructuralU32 | BenchEncoding::StructuralSparse => {
try_variable_packed_struct_per_value(Arc::new(self.clone()), field, data)?
}
BenchEncoding::Array => unreachable!(),
Expand All @@ -142,16 +154,21 @@ impl CompressionStrategy for BenchCompressionStrategy {
data: &DataBlock,
) -> Result<Box<dyn BlockCompressor>> {
let params = self.field_params(field);
if self.encoding == BenchEncoding::StructuralU32
&& let Some(compressor) = try_fixed_u8_rle_block(data, &params)?
{
let rle = match self.encoding {
BenchEncoding::Array | BenchEncoding::StructuralU16 => None,
BenchEncoding::StructuralU32 => try_fixed_u8_rle_block(data, &params)?,
BenchEncoding::StructuralSparse => try_variable_rle_block(data, &params)?,
};
if let Some(compressor) = rle {
return Ok(compressor);
}
if let Some(compressor) = try_bitpacking_block(data) {
return Ok(compressor);
}
if self.encoding == BenchEncoding::StructuralU32
&& let Some(compressor) = try_general_block(data, &params)?
if matches!(
self.encoding,
BenchEncoding::StructuralU32 | BenchEncoding::StructuralSparse
) && let Some(compressor) = try_general_block(data, &params)?
{
return Ok(compressor);
}
Expand Down Expand Up @@ -186,9 +203,11 @@ impl FieldEncodingStrategy for BenchFieldEncodingStrategy {
{
return Ok(encoder);
}
if self.encoding == BenchEncoding::StructuralU32
&& let Some(encoder) =
try_create_structural_blob(&self.primitive, field, column_index, context)?
if matches!(
self.encoding,
BenchEncoding::StructuralU32 | BenchEncoding::StructuralSparse
) && let Some(encoder) =
try_create_structural_blob(&self.primitive, field, column_index, context)?
{
return Ok(encoder);
}
Expand All @@ -202,7 +221,10 @@ impl FieldEncodingStrategy for BenchFieldEncodingStrategy {
.into(),
));
}
if self.encoding == BenchEncoding::StructuralU32 {
if matches!(
self.encoding,
BenchEncoding::StructuralU32 | BenchEncoding::StructuralSparse
) {
if let Some(encoder) = try_create_map(field, column_index, context)? {
return Ok(encoder);
}
Expand Down Expand Up @@ -268,6 +290,11 @@ pub fn encoding_strategy(encoding: BenchEncoding) -> Box<dyn FieldEncodingStrate
PrimitivePageEncoding::constant(),
PrimitivePageEncoding::dense_u32(compression),
],
BenchEncoding::StructuralSparse => vec![
PrimitivePageEncoding::sparse(compression.clone()),
PrimitivePageEncoding::constant(),
PrimitivePageEncoding::dense_u32(compression),
],
BenchEncoding::Array => unreachable!(),
};
Box::new(BenchFieldEncodingStrategy {
Expand Down
149 changes: 143 additions & 6 deletions rust/lance-encoding/benches/decoder.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
// SPDX-FileCopyrightText: Copyright The Lance Authors
use std::{collections::HashMap, hint::black_box, sync::Arc};

use arrow_array::{RecordBatch, UInt32Array};
use arrow_array::{RecordBatch, StringArray, UInt32Array};
#[cfg(feature = "bitpacking")]
use arrow_buffer::ArrowNativeType;
use arrow_schema::{DataType, Field, Schema, TimeUnit};
Expand All @@ -24,6 +24,7 @@ use lance_encoding::data::{BlockInfo, DataBlock, FixedWidthDataBlock};
#[cfg(feature = "bitpacking")]
use lance_encoding::encodings::physical::bitpacking::{ELEMS_PER_CHUNK, InlineBitpacking};
use lance_encoding::{
BufferScheduler, EncodingsIo,
decoder::{
DecodeBatchScheduler, DecoderConfig, DecoderPlugins, EncodedBatchLayout, FilterExpression,
create_decode_stream,
Expand Down Expand Up @@ -69,9 +70,9 @@ const PRIMITIVE_TYPES_FOR_FSL: &[DataType] = &[DataType::Int8, DataType::Float32
fn encoded_batch_layout(encoding: BenchEncoding) -> EncodedBatchLayout {
match encoding {
BenchEncoding::Array => EncodedBatchLayout::Array,
BenchEncoding::StructuralU16 | BenchEncoding::StructuralU32 => {
EncodedBatchLayout::Structural
}
BenchEncoding::StructuralU16
| BenchEncoding::StructuralU32
| BenchEncoding::StructuralSparse => EncodedBatchLayout::Structural,
}
}

Expand Down Expand Up @@ -589,6 +590,140 @@ fn bench_decode_compressed_parallel(c: &mut Criterion) {
}
}

async fn decode_take(
encoded: &lance_encoding::encoder::EncodedBatch,
indices: &[u64],
cache: Arc<LanceCache>,
) -> RecordBatch {
let io_scheduler = Arc::new(BufferScheduler::new(encoded.data.clone())) as Arc<dyn EncodingsIo>;
let filter = FilterExpression::no_filter();
let mut decode_scheduler = DecodeBatchScheduler::try_new(
encoded.schema.as_ref(),
&encoded.top_level_columns,
&encoded.page_table,
&vec![],
encoded.num_rows,
Arc::<DecoderPlugins>::default(),
io_scheduler.clone(),
cache,
&filter,
&DecoderConfig::default(),
)
.await
.unwrap();
let (tx, rx) = unbounded_channel();
decode_scheduler.schedule_take(indices, &filter, tx, io_scheduler);
let mut stream = create_decode_stream(
&encoded.schema,
indices.len() as u64,
indices.len() as u32,
true,
false,
true,
rx,
None,
)
.unwrap();
stream.next().await.unwrap().task.await.unwrap()
}

fn bench_variable_offsets_decode(c: &mut Criterion) {
const NUM_ROWS: usize = 262_144;
const NUM_TAKES: usize = 512;

let metadata = HashMap::from([
(
"lance-encoding:structural-encoding".to_string(),
"miniblock".to_string(),
),
(
"lance-encoding:dict-divisor".to_string(),
"100000".to_string(),
),
("lance-encoding:compression".to_string(), "none".to_string()),
]);
let schema = Arc::new(Schema::new(vec![
Field::new("value", DataType::Utf8, false).with_metadata(metadata),
]));
let lance_schema = Arc::new(lance_core::datatypes::Schema::try_from(schema.as_ref()).unwrap());
let corpora = [
(
"range",
Arc::new(StringArray::from_iter_values(
(0..NUM_ROWS).map(|index| format!("row_{index:012}")),
)) as Arc<dyn arrow_array::Array>,
),
(
"delta",
Arc::new(StringArray::from_iter_values(
(0..NUM_ROWS).map(|index| "x".repeat(4 + index % 64)),
)) as Arc<dyn arrow_array::Array>,
),
];
let mut take_indices = (0..NUM_TAKES)
.map(|index| (index as u64).wrapping_mul(104_729).wrapping_add(8_191) % NUM_ROWS as u64)
.collect::<Vec<_>>();
take_indices.sort_unstable();
take_indices.dedup();

let rt = tokio::runtime::Runtime::new().unwrap();
let mut group = c.benchmark_group("variable_offsets_decode");

for (workload, array) in corpora {
let batch = RecordBatch::try_new(schema.clone(), vec![array]).unwrap();
for (version, encoding) in [
("v2.2", BenchEncoding::StructuralU32),
("v2.3", BenchEncoding::StructuralSparse),
] {
let strategy = encoding_strategy(encoding);
let options = EncodingOptions::default();
let encoded = rt
.block_on(encode_batch(
&batch,
lance_schema.clone(),
strategy.as_ref(),
&options,
))
.unwrap();
let encoded_bytes = encoded.data.len();
let benchmark_suffix = format!("{workload}/{version}/{encoded_bytes}B");

group.throughput(criterion::Throughput::Elements(NUM_ROWS as u64));
group.bench_function(format!("scan/{benchmark_suffix}"), |bencher| {
bencher.iter(|| {
let decoded = rt
.block_on(lance_encoding::decoder::decode_batch(
&encoded,
&FilterExpression::no_filter(),
Arc::<DecoderPlugins>::default(),
false,
encoded_batch_layout(encoding),
Some(Arc::new(LanceCache::no_cache())),
))
.unwrap();
assert_eq!(decoded.num_rows(), NUM_ROWS);
})
});

for cache_mode in ["cold", "warm"] {
let cache = if cache_mode == "cold" {
Arc::new(LanceCache::no_cache())
} else {
Arc::new(LanceCache::with_capacity(64 * 1024 * 1024))
};
group.throughput(criterion::Throughput::Elements(take_indices.len() as u64));
group.bench_function(format!("take/{benchmark_suffix}/{cache_mode}"), |bencher| {
bencher.iter(|| {
let decoded =
rt.block_on(decode_take(&encoded, &take_indices, cache.clone()));
assert_eq!(decoded.num_rows(), take_indices.len());
})
});
}
}
}
}

#[cfg(feature = "bitpacking")]
fn make_inline_bitpacking_chunk<T>(bit_width: usize) -> LanceBuffer
where
Expand Down Expand Up @@ -726,13 +861,15 @@ criterion_group!(
.with_profiler(lance_testing::pprof::PProfProfiler::new(100, lance_testing::pprof::Output::Flamegraph(None)));
targets = bench_decode, bench_decode_fsl, bench_decode_str_with_dict_encoding, bench_decode_packed_struct,
bench_decode_str_with_fixed_size_binary_encoding, bench_decode_compressed,
bench_decode_compressed_parallel, bench_decode_inline_bitpacking_unchunk);
bench_decode_compressed_parallel, bench_decode_inline_bitpacking_unchunk,
bench_variable_offsets_decode);

// Non-linux version does not support pprof.
#[cfg(not(target_os = "linux"))]
criterion_group!(
name=benches;
config = Criterion::default().significance_level(0.1).sample_size(10);
targets = bench_decode, bench_decode_fsl, bench_decode_str_with_dict_encoding, bench_decode_packed_struct,
bench_decode_compressed, bench_decode_compressed_parallel, bench_decode_inline_bitpacking_unchunk);
bench_decode_compressed, bench_decode_compressed_parallel, bench_decode_inline_bitpacking_unchunk,
bench_variable_offsets_decode);
criterion_main!(benches);
Loading
Loading