pycudacov 0.0.32

A PyCuda Covariance Matrix Parallel Implementation

PyCUDACov - A PyCuda Covariance Matrix Parallel Implementation

Usage and Installation

Requires CUDA enviroment.

Installation:

$ pip install pycudacov

Basic Usage

from sklearn.datasets import make_blobs
from sklearn.preprocessing import StandardScaler
from pandas import DataFrame
import numpy as np
from pycudacov import get_cov

# Generate test dataset
rows, cols = 16384, 1024 # samples, features
X, y = make_blobs(n_samples = rows, centers = 2, n_features = cols)
X_std = StandardScaler().fit_transform(X) # Optional
df = DataFrame(X_std)
df = df.astype(np.float32)

# Call to PyCUDA Kernel, return the cov. matrix and
# GPU execution time in milliseconds
covariance_matrix, gpu_exec_time = get_cov(df.values)

Limitations

-The maximum number of features or columns of the data matrix is up to 1024

License

MIT