split benches

Signed-off-by: Onur Satici <[email protected]>

Author: onursatici

encodings/fsst/Cargo.toml

@@ -44,6 +44,6 @@ name = "fsst_compress"
 harness = false
 
 [[bench]]
-name = "chunked_dict_array_builder"
+name = "chunked_dict_fsst_builder"
 harness = false
 required-features = ["test-harness"]

…st/benches/chunked_dict_array_builder.rs …sst/benches/chunked_dict_fsst_builder.rsencodings/fsst/benches/chunked_dict_array_builder.rs renamed to encodings/fsst/benches/chunked_dict_fsst_builder.rs encodings/fsst/benches/chunked_dict_array_builder.rs renamed to encodings/fsst/benches/chunked_dict_fsst_builder.rs

@@ -2,9 +2,7 @@
 // SPDX-FileCopyrightText: Copyright the Vortex contributors
 
 use divan::Bencher;
-use rand::distr::{Distribution, StandardUniform};
 use vortex_array::arrays::ChunkedArray;
-use vortex_array::arrays::dict_test::gen_dict_primitive_chunks;
 use vortex_array::builders::builder_with_capacity;
 use vortex_array::compute::warm_up_vtables;
 use vortex_array::{Array, ArrayRef, IntoArray};
@@ -25,36 +23,6 @@ const BENCH_ARGS: &[(usize, usize, usize)] = &[
     (1000, 1000, 100),
 ];
 
-#[divan::bench(types = [u32, u64, f32, f64], args = BENCH_ARGS)]
-fn chunked_dict_primitive_canonical_into<T: NativePType>(
-    bencher: Bencher,
-    (len, unique_values, chunk_count): (usize, usize, usize),
-) where
-    StandardUniform: Distribution<T>,
-{
-    let chunk = gen_dict_primitive_chunks::<T, u16>(len, unique_values, chunk_count);
-
-    bencher.with_inputs(|| chunk.clone()).bench_values(|chunk| {
-        let mut builder = builder_with_capacity(chunk.dtype(), len * chunk_count);
-        chunk.append_to_builder(builder.as_mut());
-        builder.finish()
-    })
-}
-
-#[divan::bench(types = [u32, u64, f32, f64], args = BENCH_ARGS)]
-fn chunked_dict_primitive_into_canonical<T: NativePType>(
-    bencher: Bencher,
-    (len, unique_values, chunk_count): (usize, usize, usize),
-) where
-    StandardUniform: Distribution<T>,
-{
-    let chunk = gen_dict_primitive_chunks::<T, u16>(len, unique_values, chunk_count);
-
-    bencher
-        .with_inputs(|| chunk.clone())
-        .bench_values(|chunk| chunk.to_canonical())
-}
-
 fn make_dict_fsst_chunks<T: NativePType>(
     len: usize,
     unique_values: usize,

vortex-array/Cargo.toml

@@ -140,3 +140,8 @@ harness = false
 name = "expr_large_struct_pack"
 path = "benches/expr/large_struct_pack.rs"
 harness = false
+
+[[bench]]
+name = "chunked_dict_builder"
+harness = false
+required-features = ["test-harness"]

vortex-array/benches/chunked_dict_builder.rs

@@ -0,0 +1,54 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright the Vortex contributors
+
+use divan::Bencher;
+use rand::distr::{Distribution, StandardUniform};
+use vortex_array::Array;
+use vortex_array::arrays::dict_test::gen_dict_primitive_chunks;
+use vortex_array::builders::builder_with_capacity;
+use vortex_array::compute::warm_up_vtables;
+use vortex_dtype::NativePType;
+
+fn main() {
+    warm_up_vtables();
+    divan::main();
+}
+
+const BENCH_ARGS: &[(usize, usize, usize)] = &[
+    (1000, 10, 10),
+    (1000, 100, 10),
+    (1000, 1000, 10),
+    (1000, 10, 100),
+    (1000, 100, 100),
+    (1000, 1000, 100),
+];
+
+#[divan::bench(types = [u32, u64, f32, f64], args = BENCH_ARGS)]
+fn chunked_dict_primitive_canonical_into<T: NativePType>(
+    bencher: Bencher,
+    (len, unique_values, chunk_count): (usize, usize, usize),
+) where
+    StandardUniform: Distribution<T>,
+{
+    let chunk = gen_dict_primitive_chunks::<T, u16>(len, unique_values, chunk_count);
+
+    bencher.with_inputs(|| chunk.clone()).bench_values(|chunk| {
+        let mut builder = builder_with_capacity(chunk.dtype(), len * chunk_count);
+        chunk.append_to_builder(builder.as_mut());
+        builder.finish()
+    })
+}
+
+#[divan::bench(types = [u32, u64, f32, f64], args = BENCH_ARGS)]
+fn chunked_dict_primitive_into_canonical<T: NativePType>(
+    bencher: Bencher,
+    (len, unique_values, chunk_count): (usize, usize, usize),
+) where
+    StandardUniform: Distribution<T>,
+{
+    let chunk = gen_dict_primitive_chunks::<T, u16>(len, unique_values, chunk_count);
+
+    bencher
+        .with_inputs(|| chunk.clone())
+        .bench_values(|chunk| chunk.to_canonical())
+}

vortex-duckdb/src/exporter/dict.rs

@@ -7,8 +7,7 @@ use std::sync::Arc;
 use bitvec::macros::internal::funty::Fundamental;
 use num_traits::AsPrimitive;
 use parking_lot::Mutex;
-use vortex::arrays::DictArray;
-use vortex::arrays::{ConstantArray, ConstantVTable, PrimitiveArray};
+use vortex::arrays::{ConstantArray, ConstantVTable, DictArray, PrimitiveArray};
 use vortex::dtype::{IntegerPType, match_each_integer_ptype};
 use vortex::error::VortexResult;
 use vortex::mask::Mask;
@@ -149,8 +148,7 @@ impl<I: IntegerPType + AsPrimitive<u32>> ColumnExporter for DictExporter<I> {
 #[cfg(test)]
 mod tests {
     use vortex::IntoArray;
-    use vortex::arrays::DictArray;
-    use vortex::arrays::{ConstantArray, PrimitiveArray};
+    use vortex::arrays::{ConstantArray, DictArray, PrimitiveArray};
     use vortex::buffer::Buffer;
     use vortex::error::VortexResult;
 

vortex-duckdb/src/exporter/mod.rs

@@ -24,8 +24,7 @@ use bitvec::view::BitView;
 pub use cache::ConversionCache;
 pub use decimal::precision_to_duckdb_storage_size;
 use itertools::Itertools;
-use vortex::arrays::DictVTable;
-use vortex::arrays::{ConstantVTable, StructArray, TemporalArray};
+use vortex::arrays::{ConstantVTable, DictVTable, StructArray, TemporalArray};
 use vortex::dtype::datetime::is_temporal_ext_type;
 use vortex::encodings::runend::RunEndVTable;
 use vortex::encodings::sequence::SequenceVTable;

vortex-duckdb/src/exporter/struct_.rs

@@ -54,8 +54,7 @@ mod tests {
     use std::ffi::CString;
 
     use vortex::IntoArray;
-    use vortex::arrays::DictArray;
-    use vortex::arrays::{ConstantArray, PrimitiveArray, VarBinViewArray};
+    use vortex::arrays::{ConstantArray, DictArray, PrimitiveArray, VarBinViewArray};
     use vortex::buffer::{BitBuffer, buffer};
     use vortex::error::{VortexExpect, VortexUnwrap};
     use vortex::validity::Validity;

vortex-file/src/tests.rs

@@ -10,10 +10,9 @@ use futures::{StreamExt, TryStreamExt, pin_mut};
 use itertools::Itertools;
 use vortex_array::accessor::ArrayAccessor;
 use vortex_array::arrays::{
-    ChunkedArray, ConstantArray, DecimalArray, ListArray, PrimitiveArray, StructArray, VarBinArray,
-    VarBinViewArray,
+    ChunkedArray, ConstantArray, DecimalArray, DictEncoding, DictVTable, ListArray, PrimitiveArray,
+    StructArray, VarBinArray, VarBinViewArray,
 };
-use vortex_array::arrays::{DictEncoding, DictVTable};
 use vortex_array::expr::{
     Pack, PackOptions, VTableExt, and, eq, get_item, gt, gt_eq, lit, lt, lt_eq, or, root, select,
 };

vortex-python/src/arrays/native.rs

@@ -6,10 +6,10 @@ use std::ops::Deref;
 use pyo3::PyClass;
 use pyo3::exceptions::PyTypeError;
 use pyo3::prelude::*;
-use vortex::arrays::DictVTable;
 use vortex::arrays::{
-    BoolVTable, ChunkedVTable, ConstantVTable, DecimalVTable, ExtensionVTable, FixedSizeListVTable,
-    ListVTable, NullVTable, PrimitiveVTable, StructVTable, VarBinVTable, VarBinViewVTable,
+    BoolVTable, ChunkedVTable, ConstantVTable, DecimalVTable, DictVTable, ExtensionVTable,
+    FixedSizeListVTable, ListVTable, NullVTable, PrimitiveVTable, StructVTable, VarBinVTable,
+    VarBinViewVTable,
 };
 use vortex::encodings::alp::{ALPRDVTable, ALPVTable};
 use vortex::encodings::bytebool::ByteBoolVTable;