A powerful implementation of Incremental Principal Components Analysis that runs on GPU, built on top of PyTorch. Up to 20x speed-ups on gigabyte-scale PCA.
pip install PCAonGPU
- Clone the repository:
git clone https://github.com/dnhkng/PCAonGPU.git
- Navigate to the cloned directory:
cd PCAonGPU
- Install the required dependencies:
pip install -r requirements.txt
from gpu_pca import IncrementalPCAonGPU
# Create an instance
model = IncrementalPCAonGPU(n_components=5)
# Fit the model (either using `fit` or `partial_fit`)
model.fit(your_data)
# Transform the data
transformed_data = model.transform(your_data)SKlearn on an AMD Ryzen 9 5900X 12-Core Processor vs PCAonGPU on an Nvidia 4090
Data size: 5000 samples of 5000 dimensional data:
> python tests/benchmark_gpu_pca.py
test_sklearn_pca took 21.78324556350708 seconds to complete its execution.
test_gpu_pca took 6.523377895355225 seconds to complete its execution.
Data size: 50000 samples of 10000 dimensional data.
> python tests/benchmark_gpu_pca.py
test_sklearn_pca took 314.3792634010315 seconds to complete its execution.
test_gpu_pca took 35.23140811920166 seconds to complete its execution.