scale-lab / drills Goto Github PK

Hey,

Sorry for asking so bluntly:
I'm working on something similar using the Mockturtle library and I'm wondering where/how the ASAP7.lib file was obtained and if there is a GENLIB version of it.

thanks for your help!

Best,
Roman

Can you also add the option for a fixed optimization script to run before DRiLLS?

Hi,
Thank you for such great work. Currently, I'm trying to reproduce the result from your paper. But when I run your source code, I have noticed that there is no tech.lib and design.v file. Could you give me a link to download the sample files ?
I have tried to use the standard cell library and examples from this repository but it doesn't work. https://github.com/The-OpenROAD-Project/asap7 (error: http://i.imgur.com/foeMq5x.png)

Thank you in advance.

Issue Faced and Potential Risks for the future

• DRiLLS is currently based on tensorflow==1.12.0 , and when tried to be installed on windows with command:
  pip install tensorflow==1.12.0, returns an error of package not being found for python3.7.
• Although the support for tensorflow 1.12 for linux/debian based OS still remains, it will not be the case in the near future.

I would recommend to upgrading to tensorflow v2.

Changes to DRiLLS Framework

• The tensorflow's import statement will change as follows, to support all instances of v1 functions:
  import tensorflow.compat.v1 as tf.
• The version of tensorflow in requirements.txt will change.

I would like to take up this issue, once the Dopamine implementation is complete.

Tensorflow installed via pip uses the AVX instruction set at compile time.

To Reproduce
Steps to reproduce the behavior:

1. Clone the latest DRiLLS
2. Follow steps in setup,
3. python3.6 drills.py -help
4. Failure message

F tensorflow/core/platform/cpu_feature_guard.cc:37] The TensorFlow library was compiled to use AVX instructions, but these aren't available on your machine.
Aborted (core dumped)

Expected behavior
Expected to display drills.py help

Additional context
Found this online. The suggested fix is to get a TensorFlow that doesn't use AVX but it hasn't worked for me yet.