cuda2musa-ease

CUDAtoMUSA-Ease facilitates a seamless migration from CUDA to MUSA, highlighting MUSA's high compatibility and exceptional ease of use for developers.

NVidia Environment

1. Setup on Host A. docker pull nvidia/cuda:12.3.1-runtime-ubuntu22.04 B. docker run -it --name cuda2musa-smoothly --gpus all -v $(pwd):/host -p 80:7860 -v models:/root/.cache/ sh-harbor.mthreads.com/cuda12.3.1-runtime-ubuntu22.04 bash
2. Setup on Container A. apt install vim B. sh Anaconda3-2024.10-1-Linux-x86_64.sh -b a. /root/anaconda3/condabin/conda init b. source ~/.bashrc c. conda create --name py310 python=3.10 d. pip install -r requirements.txt
3. python gradio_app.py
4. http://{IP}