Skip to content
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

No open-cl documentation #28

Open
Whytehorse opened this issue Jan 2, 2017 · 7 comments
Open

No open-cl documentation #28

Whytehorse opened this issue Jan 2, 2017 · 7 comments

Comments

@Whytehorse
Copy link

Very confused. How does one install and test this? It's basically just a big old directory with lots of files and the readme is just a copy of the tensorflow install which tell you how to install tensorflow, not tensorflow-opencl.

@benoitsteiner
Copy link
Owner

benoitsteiner commented Jan 2, 2017

Thanks for you interest. Please take a look at the OpenCL specific section of the overall documentation. This will be published on the tensorflow.org site soon.

@atlury
Copy link

atlury commented Jan 2, 2017

Hello,
Thanks. Will wait for you guys to official open it up on tensorflow.org site. I hope intel gpus are supported as the documentation under the opencl specific section only talks about AMD

@Whytehorse
Copy link
Author

Whytehorse commented Jan 2, 2017

I looked there. I get a bunch of errors when running configure for tensorflow, like this:
ERROR: /home/ben/tensorflow-opencl-master/tensorflow/workspace.bzl:17:3: //external:eigen_archive: no such attribute 'urls' in 'new_http_archive' rule. Ok, I resolved this. It's a known bug in bazel tensorflow/tensorflow#6436

@Whytehorse
Copy link
Author

Whytehorse commented Jan 3, 2017

atlury, any gpu which has drivers that support OpenCL 1.2 will work with this tensorflow-opencl. I'm currently testing it on AMD APU called Kaveri A10-7850K, aka STEAMROLLER!!!!!!!

@Whytehorse
Copy link
Author

Whytehorse commented Jan 3, 2017

anyway, I got it to compile and install via pip3. Thank you for all of this. One more thing I think needs to be clarified is --copt=-march=native. This is supposed to optimize tensorflow to take advantage of my CPU features. But I can't find a list of those optimizations. Is there a file which lists all of these?
Found a list of gcc options here: https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html
On a AMD A10-7850k the one that worked and compiled was bdver3

@outlace
Copy link

outlace commented Apr 8, 2017

Is this going to be merged with the main tensorflow project at some point?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants