Skip to content

PythonExamples.md

Latest commit

 

History

History
53 lines (41 loc) · 1.32 KB

PythonExamples.md

File metadata and controls

53 lines (41 loc) · 1.32 KB

Using GPUZIP with Python

GPUZIP's Compressor component has been bound to Python.

The input data needs to be in cupy format, which is a NumPy-compatible format designed for CUDA. For more information, visit CuPy's official website.

A complete example can be found in the GPUZIP_Compressors repository at ./GPUZIPy/example/main.py.

Note: Currently, there are no Python bindings available for GPUZIP's Prefetch component.

Building

Dependencies (versions tested):

  • nvcc 10.1
  • CUDA 12.2
  • NVIDIA driver 536.19
  • cmake 3.22 (required)
  • make 4.2.0
  • python 3.9

Building the Python Package

git submodule update --init --recursive
cd GPUZIPy
pip install .

Running the Example

cd GPUZIPy/example
pip install -r requirements.txt
python main.py

Building with Docker

Building the Docker Image

git submodule update --init --recursive
docker build . -f ./GPUZIPy/dockerfile.gpuzipy -t maltempi/gpuzipy:latest

To specify a particular CUDA version, use the build-arg option:

git submodule update --init --recursive
docker build . -t maltempi/gpuzipy:latest --build-arg CUDA_VERSION=12.1.0

Running the Example with Docker

git submodule update --init --recursive
docker run --gpus all --rm --env NVIDIA_DISABLE_REQUIRE=1 maltempi/gpuzipy:latest