[LOGGING] Fixed LUT Sizing Issue

[AlexandreSinger]

Fixed issue where the LUT sizing for printing the circuit statistics was using the port width instead of the number of input pins.

The CLMA example from the issue:

Now looks like this:

Which makes a lot more sense.

closes #2685

EDIT:

Based on Vaughn's comments below, changed the format of the statistics to the following:

This now includes the total number of LUTs (called .names here), as well as a breakdown of their sizes.

[AlexandreSinger]

I am suprised this change failed CI. Looks like the parsing scripts may have been relying on this bug to count the number of LUTs:

I think these scripts may need to be checked. Or maybe it is as simple as updating the golden results.

[vaughnbetz]

Thanks. I suggest adding a total LUTs line as well, and then updating the parsing rule. That should fix CI (seems we were parsing the 6-LUT line as the total LUTs! ).

[AlexandreSinger]

@vaughnbetz That makes sense to me! Would you like me to include 0-LUTs in that total? I am not 100% on the semantics of what a "0-LUT" is beyond being a constant generator.

[vaughnbetz]

Sure, I'd count it. It is a constant generator.

[AlexandreSinger]

@vaughnbetz I have updated the circuit statistics to include the total number of LUTs. I have tried to follow the format of the rest of the logs:

Let me know what you think. I personally like it; but I can understand if people may be confused by ".output" coming after all the LUT sizes.

[vaughnbetz]

Looks good to me!

[AlexandreSinger]

@vaughnbetz This PR has passed CI. Please review and merge when you have a moment. I have updated the parsing script, which required me to regenerate the golden results (since some testcases were mis-counting the number of LUTs in designs).