apple / tensorflow_macos

TensorFlow for macOS 11.0+ accelerated using Apple's ML Compute framework.
Other
3.67k stars 310 forks source link

ML Compute: Cannot Set Device #256

Closed DLWCMD closed 3 years ago

DLWCMD commented 3 years ago

By following the guidance from https://github.com/apple/tensorflow_macos, I was able to create a Conda environment for the ML_Compute version of TensorFlow.

I then ran a test script (see attached) to test GPU performance, but regardless of settings (either 'gpu' or 'any'), the code is run on my CPU. Note in the script i disable eager execution, as instructed.

The image below summarizes my operating environment.

Thanks for your help.

MLC_Test_1.py.zip

Screen Shot 2021-05-09 at 9 52 11 AM
ManuelSchneid3r commented 3 years ago

235 please close this one

DLWCMD commented 3 years ago

Just to confirm, I assume you view my issue (#256) as duplicating #235. While I see the similarities, and I am am happy to accommodate your request, note that my operating environment is not M1, but rather an Intel-based MacBookPro. Implicit in my issue is whether the current release of ML Compute will support my MacBook.

Thanks for the quick response.