bit-counter 0.1.0

No project description provided

bit-counter

Package for counting the number of one bits in a numpy array of uint8 values. Implemented as a Python module using Rust, providing high performance counting.

Building

To build this package an installation of Rust and Python with the maturin package is required. The Maturin documentation on maturin local development is a useful reference for more details.

When building it is best to set the RUSTFLAGS environment variable to specify targeting CPUs with the POPCNT instruction. i.e. RUSTFLAGS='-C target-feature=+popcnt' maturin build -r.

Example usage

import numpy as np
from bit_counter import count_ones

arr = np.packbits(np.random.choice([True, False], 1000000))

count_of_true_values = count_ones(arr)

Performance

When built to target a CPU with popcnt support the count_ones method provided is substantially faster than a naive np.unpackbits(arr).sum(). The count_ones method also doesn't require unpacking the bit packed numpy array, so doesn't require any addtional memory to do the calculation.

For example with the following test code for 100 million True/False values that are packed:

import numpy as np
from bit_counter import count_ones

arr = np.packbits(np.random.choice([True, False], 100000000))

Using this package on a mid-2013 MacBook Air (1.7 GHz Intel Core i7) yields:

%timeit count_ones(arr)

45.3 ms ± 2.08 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

Compared to the simple unpack and sum case:

%timeit np.unpackbits(arr).sum()

108 ms ± 6.89 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)