Fame PyTorch ROCm Kit – Installation Instructions 🚀
This file explains how to manually install the Fame-built PyTorch ROCm version.
🛠️ 1. Prepare Environment
Activate your Python environment (recommended: venv):
source ~/rocm_env/bin/activate
Install CMake if missing:
pip install cmake
Ensure HIP and ROCm paths are correctly set (ROC_ENABLE_PRE_VEGA=1, etc).
🛠️ 2. Build PyTorch
Navigate to the Fame Kit folder:
cd fame-pytorch-kit/
Run the build script:
bash build_torch.sh
📦 3. Install the Wheel
Navigate to the dist folder:
cd pytorch/dist/
Install the Wheel manually:
pip install torch-2.8.0a0+gitc402b3b-cp312-cp312-linux_x86_64.whl
✅ 4. Verify Installation
Test your installation:
python -c "import torch; print(torch.cuda.is_available())"
python -c "import torch; print(torch.cuda.get_device_name(0))"
python -c "import torch; print(torch.version.hip)"
You should see:
True(GPU detected)- Your GPU model name
- HIP version string