oneAPI backend update: report

[enlupi]

Description

Added functionalities for parsing and printing the QoR report when using oneAPI backend.

Type of change

• New feature (non-breaking change which adds functionality)

This PR introduces the possibility of parsing and printing the oneAPI report, akin to what it is already possible to do using Vivado backend. All the necessary functions are contained in the hls4ml/report/oneapi_report.py file.

Tests

Modified test/pytest/test_report.py to test parsing and printing of the report:

• Added oneAPI backend test while manteining functionalities for Vivado
• Restructured code so that the HLS model setup is executed before the test proper: the test is marked as failed only if errors happen during the parsing and printing of report
• Restructured the test_report directory so that Vivado and oneAPI files are included in separated subdirectories
• Output project directories are now temporary and deleted after test execution

Test Configuration:

Checklist

• 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.
• My changes generate no new warnings.
• 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.

[jmitrevs]

The Vivado/Vitis backends are also integrated with the build command, so that it can return a dictionary of results. Do we want to also add such functionality?

[vloncar]

Yes, all backends do this, so we should keep consistent. Should just be a line return parse_oneapi_report(outdir) at the end of OneAPIBackend.build(), right?

[enlupi]

I added the change as suggested in the previous comments. Please let me know if further work is needed on this.

[jmitrevs]

Is this generally complete and ready for review?

[jmitrevs]

This needs documentation so that we know how to use it.

[enlupi]

Yes, this is generally complete, unless of course you have idea on how to improve it.

As for the documentation, would something simple like the one for the Quartus report work?

[jmitrevs]

What is the user interface? How is the user supposed to print the report?

[enlupi]

The user interface is the same as for the Vivado backend. With parse_vivado_report the user provides the hls project directory and gets as output a dictionary containing the relevant information. WIth print_oneapi_report the user provides the dictionary and it prints the report (as a nice html table if the code is run on a notebook, or just as a "string" table if run on the terminal)

[jmitrevs]

The only interface we have ever "documented" is read_vivado_report in the tutorial. I don't think most people know how to use this. We definitely need to improve the documentation for all the backends, not just for this one. (People were wondering if read_vivado_report changed to read_vitis_report, for example.) And I still think after hls_model.build() we should be able to say hls_model.report() or something similar. But maybe all those improvements are outside of this scope of this PR. (We should also update the oneAPI branch of the tutorial using these utilities.)