[throw14082020]

When building Tensorflow from source (because you have to, to get it working even with Rosetta 2) was frustratingly slow, and the whole computer became unusable. After more than an hour of waiting, I stopped the bazel build process. Which reminds me, I'm going trigger the build now and go for a walk...

[rjsw]

Is Bazel native or running under Rosetta2 ?

[throw14082020]

First I ran it native, and after 5+ hours, it failed right at the end. So I restarted with Rosetta2, and after 2 hours, I gave up. I just downloaded the tensorflow_macOS fork that apple made to use the M1 GPU anyway.