Install TensorFlow on 32-bit Linux
Install TensorFlow on 32-bit Linux
TL;DR — TensorFlow only ships 64-bit binaries, so I built it from source for 32-bit (i386) Linux — a hands-on tour of Bazel, cross-compilation, and the platform assumptions baked into a large C++/Python codebase.
Stack — TensorFlow, Bazel, C++, Python, Linux (i386)
Instructions to build and install TensorFlow on a 32-bit Linux system, maybe just for fun.
This one started as a “can it even be done?” experiment. The official TensorFlow packages only ship 64-bit binaries, so getting it running on 32-bit (i386) Linux meant building the whole thing from source. I worked through TensorFlow’s Bazel build, fought past the places where it quietly assumes a 64-bit platform, and wrote down the toolchain, flags, and Python configuration that finally produced a working package on 32-bit hardware. The payoff is being able to bring modern machine-learning tooling back to the older, low-power machines everyone else has written off — and, just as much, it turned into a hands-on tour of how TensorFlow is actually compiled under the hood. If you want to try it yourself, the full build and install steps are here:
