Add option to configure IP version

AdrianAlan commented 1 year ago

A small change of adding an option to specify the IP version, defaulting to 1.0.0.

Motivation: although it is a long time since the Y2K22 issue I hear people still run into it when running on systems where admins didn't install the patch. The same workaround was suggested in here. Issue mentioned in #633.

Type of change

[x] Bug fix (non-breaking change that fixes an issue)
[x] New feature (non-breaking change which adds functionality)

Tests

Please suggest tests for this change.

Checklist

[x] I have read the guidelines for contributing.
[ ] I have commented my code, particularly in hard-to-understand areas.
[ ] I have made corresponding changes to the documentation.
[x] My changes generate no new warnings.
[x] I have installed and run pre-commit on the files I edited or added.
[ ] I have added tests that prove my fix is effective or that my feature works.

vloncar commented 1 year ago

Nice!

But how would one set a version using the public API? The only way to use it is to create the config manually and call the keras_to_hls() or set it after the ModelGraph was created but before the write() is called.

AdrianAlan commented 1 year ago

Now it should be possible to specify it through hls4ml.converters.create_config as well. Is there another way you thought about it?

PS Wouldn't you have the same concern about ClockUncertainty?

vloncar commented 1 year ago

Looks good. It failed for the accelerator backend, so I added support for it as well.

fastmachinelearning / hls4ml