Analyzing the runtime, energy usage, and performance of Tensorflow training on a M1 Mac Mini and Nvidia V100
An interesting comparison of Apple Intel and M1 chips machines with Nvidia 100 when using Tensorflow.
We ran a sweep of 8 different configurations of our training script and show that the Apple M1 offers impressive performance within reach of much more expensive and less energy efficient accelerators such as the Nvidia V100 for smaller architectures and datasets.
Code is available on Colab https://colab.research.google.com/drive/1RvZBpzJRW9MNPWQ2rZG8HIyHJbaCwnTI.
They also include tips on setting up a Mac mini to run Tensorflow.
Only initial results on modest data sets, will be interesting to see the performance when Apple releases more Pro hardware.