Skip to content

Docker-based environments for building Tensorflow with support for x86/amd64 (down to Celeron without AVX), ARM32v7 and ARM64v8 CPUs

License

Notifications You must be signed in to change notification settings

photonixapp/tensorflow-builder

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

11 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

tensorflow-builder

Docker-based build environment that creates Tensorflow packages that don't use optimised CPU instructions such as AVX, AVX2, SSE4.1, SSE4.2 and FMA. This is the version of Tensorflow that is used in Photonix to maintain compatibility with as many people's machines as possible. Performance testing detailed here showed a 13% longer execution time than the official, optimised builds.

Building

Running the commands below should build Python Wheel packages for each supported architecture in the wheels directory. Beware that building will probably take many hours. We usually create the largest possible VPS on DIgitalOcean with 32 CPU cores and it can build each architecture in about an hour. Check out the releases page to make use of our builds.

./build.sh amd64
./build.sh arm32v7
./build.sh arm64v8

Todo

  • Provide options for passing in custom PyPI server credentials and having wheels automatically uploaded using pypi-uploader

About

Docker-based environments for building Tensorflow with support for x86/amd64 (down to Celeron without AVX), ARM32v7 and ARM64v8 CPUs

Topics

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages