Installation

Step-by-Step Guide

GEM-pRF requires an NVIDIA GPU and CUDA for accelerated pRF computation. Make sure your system has a compatible GPU available.

Step 1. Create a new python environment (recommended)

• Create a fresh conda environment:

  conda create --name gemprf python=3.10
  conda activate gemprf

Step 2. Set up GPU environment

• Install the CUDA SDK.
• Check your CUDA installation:

  nvcc --version
  nvidia-smi

• Install a CuPy build that matches your CUDA version:

  pip install cupy-cuda12x   # Replace 12x with your CUDA version

Step 3. Verify CUDA–CuPy compatibility

• Run a quick test to confirm CuPy can use the CUDA runtime correctly:

  python -c "import cupy as cp; print(cp.arange(5) ** 2)"

• If this fails, your CuPy build does not match your CUDA runtime.

Step 4. Install GEM-pRF via pip

• Install from PyPI:

  pip install gemprf

• Latest releases: PyPI – gemprf

Step 5. Try the GEMpRF-DemoKit (optional)

• Follow the instructions in the Getting Started with GEMpRF-DemoKit tutorial to verify your installation.