Iot Columnar Bench

1import { bench, describe } from "vitest";
2import { query } from "./clickhouse.js";
3import { Cursor } from "../src/readers/core.js";
4import {
5  type IotRow,
6  decodeIotColumnar,
7  iotRowAt,
8  readIotRowFast,
9} from "../src/examples/iot.js";
10
11/**
12 * Benchmark: row-objects (AoS, `readIotRowFast`) vs columnar (SoA,
13 * `decodeIotColumnar`) over the same fixed-width IoT buffer. Columnar removes
14 * the per-row object + `Date` allocation that dominates a numeric decode, for a
15 * ~4x win — the "free 4x, in plain JS" the WASM investigation surfaced (see
16 * `case-studies/wasm-vs-js.md`).
17 */
18const N = 50_000;
19const SELECT =
20  `SELECT toUInt32(number % 1000) AS sensor_id, ` +
21  `toDateTime64(1700000000 + number, 3) AS ts, ` +
22  `20 + (number % 1500) / 100 AS temperature, ` +
23  `30 + (number % 7000) / 100 AS humidity, ` +
24  `980 + (number % 6000) / 100 AS pressure, ` +
25  `toFloat32(3 + (number % 200) / 100) AS battery, ` +
26  `toUInt8(number % 4) AS status ` +
27  `FROM numbers(${N})`;
28const BUF = await query(`${SELECT} FORMAT RowBinary`);
29
30function decodeRows(): IotRow[] {
31  const s = new Cursor(BUF);
32  const out: IotRow[] = [];
33  while (s.pos < s.buf.length) out.push(readIotRowFast(s));
34  return out;
35}
36
37// correctness: columnar (via the lazy row accessor) must equal the row decode
38{
39  const rows = decodeRows();
40  const cols = decodeIotColumnar(BUF);
41  if (rows.length !== N || cols.sensor_id.length !== N)
42    throw new Error("columnar: row count");
43  for (const i of [0, 1, 123, 4999, N - 1]) {
44    if (JSON.stringify(rows[i]) !== JSON.stringify(iotRowAt(cols, i))) {
45      throw new Error(`columnar: row ${i} mismatch`);
46    }
47  }
48}
49
50describe("IoT decode: row-objects vs columnar", () => {
51  bench("rows — readIotRowFast (objects + Date)", () => {
52    decodeRows();
53  });
54  bench("columnar — decodeIotColumnar (typed arrays)", () => {
55    decodeIotColumnar(BUF);
56  });
57});