RE: Tensorflow, Bazel, and Bazelisk: armhf on BBAI? about bazelisk HOT 2 CLOSED

Hi Seth,

are you building directly on your BBAI? If I read this correctly, it has a dual-core CPU and 1GB of RAM, right? The TensorFlow build is quite heavy, I'm not sure if we can get this to work...

It might be easier to build on a machine or VM with a compatible architecture, but more CPU power and RAM. Ideally, TensorFlow would offer prebuilt binaries for these platforms, as it is very unrealistic to build it directly on them. Even if it would work, the build would probably take an entire day. Have you asked on the TensorFlow issue tracker about this already?

But regarding your build, I assume you already managed to get Bazel running - if no, let me know and we can try to get that done first. Once you have a version of Bazel that's compatible with TensorFlow running, I would try to run the TF build using bazel build --jobs=1 to disable parallelism, as otherwise we'll run out of RAM. If this still gives you an error, could you post a longer log please?

from bazelisk.

Hello,

Yes. I was trying to build it on the BBAI. Thank you but I think I have moved on for now. I might bring up another issue later.

Seth

from bazelisk.