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

How to run those python Codes to generate verilog codes step by step? #1

Open
pualdelis opened this issue Jul 26, 2018 · 5 comments
Open

Comments

@pualdelis
Copy link

Hi, I'm interested in your projects, but I don't konw how to use those python codes, any instuctions? thanks

@ZFTurbo
Copy link
Owner

ZFTurbo commented Jul 26, 2018

If you need just veilog to exactly recreate device, then you can check verilog folder it has everything you need. In case you want to generate everything from scratch look at the "How to run" section. There are 5 scripts you need to run one by one. You must have "Python 3.5, Tensorflow >=1.4.0, Keras >=2.1.3" installed.

@pualdelis
Copy link
Author

Thanks a lot. In fact, i found that those python codes generate verilog codes only by write to file and use text replacement. In this case, may i understand like this, the work of python is to train the network and generate a text file of verilog codes by using foregoing parameters, and without really understanding it's meaning?
lastly, have those verilog CNN codes being trained and i can use it directly?
How did you come up with this brilliant idea, is there any book or paper you refered to ?

@ZFTurbo
Copy link
Owner

ZFTurbo commented Aug 14, 2018

Python code mostly used to obtain weights for neural net, then find how much precision needed for fix point representation. Next generate verilog text file for neural net with calculated precision and required number of convolution blocks. The last part with verilog generation actually written not very flexible currently.

No, there was no paper. We wrote our own paper, but it still under rewiew for publication.

@pualdelis
Copy link
Author

thanks a lot,if your paper is published, please let me know. Thanks again

@ZFTurbo
Copy link
Owner

ZFTurbo commented Aug 30, 2018

We put preprint on arxiv: https://arxiv.org/abs/1808.09945

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

2 participants