|
| 1 | +// SPDX-License-Identifier: Apache-2.0 |
| 2 | +// SPDX-FileCopyrightText: Copyright the Vortex contributors |
| 3 | + |
| 4 | +#![expect( |
| 5 | + clippy::unwrap_used, |
| 6 | + clippy::clone_on_ref_ptr, |
| 7 | + clippy::cloned_ref_to_slice_refs, |
| 8 | + clippy::cast_possible_truncation, |
| 9 | + clippy::cast_possible_wrap, |
| 10 | + clippy::redundant_clone |
| 11 | +)] |
| 12 | + |
| 13 | +//! Row-encode throughput benchmarks comparing `arrow-row` against vortex's `convert_columns` |
| 14 | +//! for the canonical scenarios shipped in PR 1: a primitive i64 column, a Utf8 column, |
| 15 | +//! and a mixed-field struct. Per-encoding fast paths (Constant, Dict, Patched, BitPacked, |
| 16 | +//! FoR, Delta) gain their own triplets in PR 3. |
| 17 | +
|
| 18 | +use std::sync::Arc; |
| 19 | + |
| 20 | +use arrow_array::Int64Array; |
| 21 | +use arrow_array::StringArray; |
| 22 | +use arrow_array::StructArray as ArrowStructArray; |
| 23 | +use arrow_row::RowConverter; |
| 24 | +use arrow_row::SortField as ArrowSortField; |
| 25 | +use arrow_schema::DataType; |
| 26 | +use arrow_schema::Field; |
| 27 | +use divan::counter::BytesCount; |
| 28 | +use mimalloc::MiMalloc; |
| 29 | +use rand::RngExt; |
| 30 | +use rand::SeedableRng; |
| 31 | +use rand::distr::Alphanumeric; |
| 32 | +use rand::rngs::StdRng; |
| 33 | +use vortex_array::IntoArray; |
| 34 | +use vortex_array::LEGACY_SESSION; |
| 35 | +use vortex_array::VortexSessionExecute; |
| 36 | +use vortex_array::arrays::PrimitiveArray; |
| 37 | +use vortex_array::arrays::StructArray; |
| 38 | +use vortex_array::arrays::VarBinViewArray; |
| 39 | +use vortex_row::SortField; |
| 40 | +use vortex_row::convert_columns; |
| 41 | + |
| 42 | +#[global_allocator] |
| 43 | +static GLOBAL: MiMalloc = MiMalloc; |
| 44 | + |
| 45 | +const N: usize = 100_000; |
| 46 | + |
| 47 | +fn main() { |
| 48 | + divan::main(); |
| 49 | +} |
| 50 | + |
| 51 | +fn gen_i64(n: usize, seed: u64) -> Vec<i64> { |
| 52 | + let mut rng = StdRng::seed_from_u64(seed); |
| 53 | + (0..n) |
| 54 | + .map(|_| rng.random_range(i64::MIN..i64::MAX)) |
| 55 | + .collect() |
| 56 | +} |
| 57 | + |
| 58 | +fn gen_words(n: usize, mean_len: usize, seed: u64) -> Vec<String> { |
| 59 | + let rng = &mut StdRng::seed_from_u64(seed); |
| 60 | + (0..n) |
| 61 | + .map(|_| { |
| 62 | + let len = rng.random_range(mean_len.saturating_sub(4)..=mean_len + 4); |
| 63 | + rng.sample_iter(&Alphanumeric) |
| 64 | + .take(len) |
| 65 | + .map(char::from) |
| 66 | + .collect::<String>() |
| 67 | + }) |
| 68 | + .collect() |
| 69 | +} |
| 70 | + |
| 71 | +// ---------- primitive_i64 ---------- |
| 72 | + |
| 73 | +#[divan::bench] |
| 74 | +fn primitive_i64_arrow_row(bencher: divan::Bencher) { |
| 75 | + let v = gen_i64(N, 0); |
| 76 | + let arr = Arc::new(Int64Array::from(v.clone())) as arrow_array::ArrayRef; |
| 77 | + let conv = RowConverter::new(vec![ArrowSortField::new(DataType::Int64)]).unwrap(); |
| 78 | + let bytes = (N * (1 + 8)) as u64; |
| 79 | + bencher |
| 80 | + .counter(BytesCount::new(bytes)) |
| 81 | + .bench_local(|| conv.convert_columns(&[arr.clone()]).unwrap()) |
| 82 | +} |
| 83 | + |
| 84 | +#[divan::bench] |
| 85 | +fn primitive_i64_vortex(bencher: divan::Bencher) { |
| 86 | + let v = gen_i64(N, 0); |
| 87 | + let col = PrimitiveArray::from_iter(v.clone()).into_array(); |
| 88 | + let bytes = (N * (1 + 8)) as u64; |
| 89 | + bencher.counter(BytesCount::new(bytes)).bench_local(|| { |
| 90 | + let mut ctx = LEGACY_SESSION.create_execution_ctx(); |
| 91 | + convert_columns(&[col.clone()], &[SortField::default()], &mut ctx).unwrap() |
| 92 | + }) |
| 93 | +} |
| 94 | + |
| 95 | +// ---------- utf8 ---------- |
| 96 | + |
| 97 | +#[divan::bench] |
| 98 | +fn utf8_arrow_row(bencher: divan::Bencher) { |
| 99 | + let words = gen_words(N, 16, 7); |
| 100 | + let total: u64 = words |
| 101 | + .iter() |
| 102 | + .map(|w| 1 + (w.len().div_ceil(32) * 33) as u64) |
| 103 | + .sum(); |
| 104 | + let arr = Arc::new(StringArray::from(words.clone())) as arrow_array::ArrayRef; |
| 105 | + let conv = RowConverter::new(vec![ArrowSortField::new(DataType::Utf8)]).unwrap(); |
| 106 | + bencher |
| 107 | + .counter(BytesCount::new(total)) |
| 108 | + .bench_local(|| conv.convert_columns(&[arr.clone()]).unwrap()) |
| 109 | +} |
| 110 | + |
| 111 | +#[divan::bench] |
| 112 | +fn utf8_vortex(bencher: divan::Bencher) { |
| 113 | + let words = gen_words(N, 16, 7); |
| 114 | + let total: u64 = words |
| 115 | + .iter() |
| 116 | + .map(|w| 1 + (w.len().div_ceil(32) * 33) as u64) |
| 117 | + .sum(); |
| 118 | + let col = VarBinViewArray::from_iter_str(words.iter().map(String::as_str)).into_array(); |
| 119 | + bencher.counter(BytesCount::new(total)).bench_local(|| { |
| 120 | + let mut ctx = LEGACY_SESSION.create_execution_ctx(); |
| 121 | + convert_columns(&[col.clone()], &[SortField::default()], &mut ctx).unwrap() |
| 122 | + }) |
| 123 | +} |
| 124 | + |
| 125 | +// ---------- struct_mixed ---------- |
| 126 | + |
| 127 | +fn struct_mixed_inputs() -> (Vec<i64>, Vec<String>, u64) { |
| 128 | + let ids = gen_i64(N, 1); |
| 129 | + let names = gen_words(N, 16, 2); |
| 130 | + // sentinel (1) + i64 (1+8=9) + utf8-name (1 + ceil(len/32)*33) |
| 131 | + let total: u64 = (0..N) |
| 132 | + .map(|i| { |
| 133 | + let name_bytes = 1 + (names[i].len().div_ceil(32) * 33) as u64; |
| 134 | + 1u64 + 9u64 + name_bytes |
| 135 | + }) |
| 136 | + .sum(); |
| 137 | + (ids, names, total) |
| 138 | +} |
| 139 | + |
| 140 | +#[divan::bench] |
| 141 | +fn struct_mixed_arrow_row(bencher: divan::Bencher) { |
| 142 | + let (ids, names, total) = struct_mixed_inputs(); |
| 143 | + let id_arr = Arc::new(Int64Array::from(ids)) as arrow_array::ArrayRef; |
| 144 | + let name_arr = Arc::new(StringArray::from(names)) as arrow_array::ArrayRef; |
| 145 | + let arrow_struct = Arc::new(ArrowStructArray::from(vec![ |
| 146 | + (Arc::new(Field::new("id", DataType::Int64, false)), id_arr), |
| 147 | + ( |
| 148 | + Arc::new(Field::new("name", DataType::Utf8, false)), |
| 149 | + name_arr, |
| 150 | + ), |
| 151 | + ])) as arrow_array::ArrayRef; |
| 152 | + let struct_fields = vec![ |
| 153 | + Arc::new(Field::new("id", DataType::Int64, false)), |
| 154 | + Arc::new(Field::new("name", DataType::Utf8, false)), |
| 155 | + ]; |
| 156 | + let conv = RowConverter::new(vec![ArrowSortField::new(DataType::Struct( |
| 157 | + struct_fields.into(), |
| 158 | + ))]) |
| 159 | + .unwrap(); |
| 160 | + bencher |
| 161 | + .counter(BytesCount::new(total)) |
| 162 | + .bench_local(|| conv.convert_columns(&[arrow_struct.clone()]).unwrap()) |
| 163 | +} |
| 164 | + |
| 165 | +#[divan::bench] |
| 166 | +fn struct_mixed_vortex(bencher: divan::Bencher) { |
| 167 | + let (ids, names, total) = struct_mixed_inputs(); |
| 168 | + let id_arr = PrimitiveArray::from_iter(ids).into_array(); |
| 169 | + let name_arr = VarBinViewArray::from_iter_str(names.iter().map(String::as_str)).into_array(); |
| 170 | + let struct_arr = StructArray::from_fields(&[("id", id_arr), ("name", name_arr)]) |
| 171 | + .unwrap() |
| 172 | + .into_array(); |
| 173 | + bencher.counter(BytesCount::new(total)).bench_local(|| { |
| 174 | + let mut ctx = LEGACY_SESSION.create_execution_ctx(); |
| 175 | + convert_columns(&[struct_arr.clone()], &[SortField::default()], &mut ctx).unwrap() |
| 176 | + }) |
| 177 | +} |
0 commit comments