-
Notifications
You must be signed in to change notification settings - Fork 64
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
How to use PlaidML backend? #129
Comments
Found this: https://ieee-region6.org/scv-cis/wp-content/uploads/sites/6/2019/03/nGraph_PlaidML.pdf and this: https://github.com/NervanaSystems/ngraph/tree/master/src/ngraph/runtime/plaidml Seems like I need some glue and an instruction manual. |
here is the good news With Ngraph
Without
|
https://github.com/tensorflow/ngraph-bridge/blob/master/python/ngraph_bridge/__init__.in.py |
First you need to build the backend as a flag tho the python script
|
Nearly there
|
|
widedeep2.zip PLAIDML:0 as backend error is as follows
|
This is what I see from the gtest
|
How do I enable NGraph for the tests? |
Trying the model_level_tests now, also seems to be CPU only |
Looking at this: How can PlaidML be specified? |
OK with the CPU back end everything is OK
see: from here:
But when I change a single line
to
I get the following in the attached output.txt file How can I diagnose what the issue is? |
I'm having the same issue with the "compilation failed" error when using the PlaidML backend. Everything works fine with the CPU backend. @samhodge: Did you ever find a solution? |
Interestingly, I get a different error when switching from Metal to OpenCL:
(I'm using a library called Ludwig, which I modified myself to work on top of TensorFlow with nGraph.) |
I havent followed up for some time, I hope to put some time into this soon Sam |
Had a similar issue to @Gerzer's:
|
To answer the original question How to use the backend. At the top of your NoteBook or python file before importing Tensorflow.
Then find
|
OK I got ngraph working with a simple model project that I am not that interested in but it is working:
see:
wide_deep.zip
tensorflow/models@b4cf230#diff-1bcc2f2c0115147dcbe54cff4c19c431
see attached zip
I see
How do I make it say using backend PlaidML , cooking your AMD GPU to infinity and beyond?
I quickly checked the docs and couldnt see anything useful.
Sam
The text was updated successfully, but these errors were encountered: