gnitz-zset 0.1.3

The Z-set kernel of the gnitz database: schema, columnar batches, cursors and operators
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94

use super::adhoc_fold::AdhocFold;
use crate::repr::{Batch, BatchBuilder};
use crate::schema::{SchemaColumn, SchemaDescriptor, TypeCode};
use gnitz_foundation::perf::Counter;
use gnitz_wire::{AggDescriptor, AggFunc, AggReadSpec};

const N: u64 = 1 << 20;
/// A store's scan chunk.
const CHUNK: u64 = 65_536;
/// A store's cap on the groups of one fold.
const GROUP_CAP: usize = 65_536;

/// Instructions per row of a whole fold — every chunk and the finish — over
/// `[U64 pk | I64 grp | I64 a | I32 b NULL | F64 c]`, across aggregate sets and
/// group counts: grouped by `grp` into a few groups and into nearly the cap, and
/// by no column, which is the global fold.
#[test]
#[ignore = "benchmark; run with --release --ignored --nocapture --test-threads=1"]
fn adhoc_fold_bench() {
    let schema = SchemaDescriptor::new(
        &[
            SchemaColumn::new(TypeCode::U64, false),
            SchemaColumn::new(TypeCode::I64, false),
            SchemaColumn::new(TypeCode::I64, false),
            SchemaColumn::new(TypeCode::I32, true),
            SchemaColumn::new(TypeCode::F64, false),
        ],
        &[0],
    );
    let mix = |i: u64| i.wrapping_mul(0x9E37_79B9_7F4A_7C15);
    let agg = |col_idx: u32, agg_op: AggFunc| AggDescriptor { col_idx, agg_op };
    let agg_sets: [(&str, Vec<AggDescriptor>); 4] = [
        ("count+sum", vec![AggDescriptor::COUNT_STAR, agg(2, AggFunc::Sum)]),
        (
            "count+3sum+cnn",
            vec![
                AggDescriptor::COUNT_STAR,
                agg(2, AggFunc::Sum),
                agg(3, AggFunc::Sum),
                agg(4, AggFunc::Sum),
                agg(3, AggFunc::CountNonNull),
            ],
        ),
        (
            "count+sum+min+max",
            vec![
                AggDescriptor::COUNT_STAR,
                agg(2, AggFunc::Sum),
                agg(2, AggFunc::Min),
                agg(3, AggFunc::Max),
            ],
        ),
        // A PK column has no column kernel: each is stepped a row at a time.
        ("pk min+sum", vec![agg(0, AggFunc::Min), agg(0, AggFunc::Sum)]),
    ];
    let counter = Counter::instructions();
    for groups in [0u64, 16, 60_000] {
        let chunks: Vec<Batch> = (0..N / CHUNK)
            .map(|c| {
                let mut b = BatchBuilder::new(&schema);
                for i in c * CHUNK..(c + 1) * CHUNK {
                    b.begin_row(i as u128, 1);
                    b.put_int((mix(i) >> 20) as u128 % groups.max(1) as u128);
                    b.put_int(mix(i ^ 7) as i64 as u128);
                    b.put_opt_int((i % 5 != 0).then(|| mix(i ^ 9) as u32 as u128));
                    b.put_float((mix(i) >> 11) as f64 / 3.0);
                    b.end_row();
                }
                b.finish()
            })
            .collect();
        let group_cols = if groups == 0 { vec![] } else { vec![1] };
        for (name, aggs) in &agg_sets {
            let spec = AggReadSpec {
                group_cols: group_cols.clone(),
                aggs: aggs.clone(),
            };
            let fold = || {
                let mut fold = AdhocFold::new(&schema, &spec, GROUP_CAP).unwrap();
                for c in &chunks {
                    fold.fold_ranges(c, &[(0, c.len())]).unwrap();
                }
                fold.finish()
            };
            std::hint::black_box(fold());
            let (out, instructions) = counter.measure(fold);
            std::hint::black_box(out);
            println!(
                "adhoc_fold_bench groups={groups:<6} {name:<18} {:>6.1} instr/row",
                instructions as f64 / N as f64
            );
        }
    }
}