1、安装TensorFlow(GPU版)
xcode-select --install
conda create -n tf-mac-gpu python=3.8
conda activate tf-mac-gpu
conda install -c apple tensorflow-deps
pip install tensorflow-macos
pip install tensorflow-metal
conda install jupyter notebook
更新包:
pip install numpy --upgrade
pip install pandas --upgrade
pip install matplotlib --upgrade
pip install scikit-learn --upgrade
pip install scipy --upgrade
pip install plotly --upgrade
2、安装pytorch(GPU版)
xcode-select --install
conda create -n torch-mac-gpu python=3.8
conda activate torch-mac-gpu
conda install pytorch torchvision torchaudio -c pytorch-nightly
conda install jupyter notebook
3、测试对比结果
3.1 TensorFlow使用kaggle上dog-vs-cat分割数据,速度是使用Mac CPU的6-7倍。
3.2 pytorch只测试了MNIST,速度是使用Mac CPU的2.5倍。
Apple给的对比数据:experiments-and-benchmarks-for-m1
4、主要参考链接:
4.1、install-tensorflow-on-mac-m1-m2-with-gpu