bench-bin search.py

import os
import random
import polars as pl
import timeit
import tqdm

results = []

for shuffle_rhs in [False]:
    for lhs_size in tqdm.tqdm([11, 100, 1_000, 10_000, 100_000, 1_000_000, 10_000_000]):
        for rhs_size in [1, 10, 100, 1_000, 10_000, 100_000, 1_000_000, 10_000_000]:
            lhs = pl.Series(range(lhs_size))
            rhs = pl.Series(range(rhs_size))
            if shuffle_rhs:
                rhs = rhs.shuffle()
            nrepeat = 5
            bench = lambda number: min(timeit.repeat('lhs.is_in(rhs)', number=number, globals=globals(), repeat=nrepeat))/number
            time1 = bench(1)
            number= min(100, max(3, int(1e-2/time1)))
            time = bench(number)
            results.append({
                "shuffled": shuffle_rhs,
                "lhs_size": lhs_size,
                "rhs_size": rhs_size,
                f"{os.getenv('bench_name')}_time": time,
            })

pl.DataFrame(results).write_csv(f"/tmp/{os.getenv('bench_name')}.csv")