update fracturable LUT connectivity based on known legal external placements

[KA7E]

Updated fracturable LUT connectivity in order to successfully reconstruct known legal clusters.

Description

Added complete interconnect from fracturable LUT data inputs to data inputs for each half LUT.

Related Issue

Motivation and Context

Unable to reconstruct known legal RippleFPGA-generated clusters without this change.

How Has This Been Tested?

Successfully reconstructed RippleFPGA solutions for all 12 ISPD benchmarks.

Types of changes

• Bug fix (change which fixes an issue)
• New feature (change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

• My change requires a change to the documentation
• I have updated the documentation accordingly
• I have added tests to cover my changes
• All new and existing tests passed

[AlexandreSinger]

@vaughnbetz One of the three PRs merged yesterday is causing vtr_reg_nightly_test5 to timout after 7 hours (it usually takes 3). I think this PR may be the culprit since this nightly test contains the vpr_ispd test and the other two PRs have nothing to do with vtr_reg_nightly_test5. Something seems to be causing these tests to take WAY longer than expected.

[KA7E]

I'll take a look.

…

On Tue, Jul 30, 2024, 09:07 AlexandreSinger ***@***.***> wrote: @vaughnbetz <https://github.com/vaughnbetz> One of the three PRs merged yesterday is causing vtr_reg_nightly_test5 to timout after 7 hours (it usually takes 3). I think this PR may be the culprit since this nightly test contains the vpr_ispd test and the other two PRs have nothing to do with vtr_reg_nightly_test5. Something seems to be causing these tests to take WAY longer than expected. — Reply to this email directly, view it on GitHub <#2668 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AM6ZD4XBVT3OKFKU3S67ZETZO6FXLAVCNFSM6AAAAABLUTSW4CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENJYGMYDQMZVGE> . You are receiving this because you authored the thread.Message ID: ***@***.*** com>

[AlexandreSinger]

@KA7E Thank you! I have raised a PR to undo this change so I can run the CI and see if it gets resolved:
#2670

I do not think it will be necessary to merge this reverting PR, but it will just verify if this is causing the issue. Perhaps this change makes the ISPD testcases a bit harder, and we may just need to move the ISPD tests into their own nightly_test.

[KA7E]

It shouldn't make them harder since it adds connectivity, but it may make packing slower? I don't mind if you change it back for now. It's only needed for legalizing ispd placer generated placements which is not something anyone but me needs to do right now.

…

On Tue, Jul 30, 2024, 09:30 AlexandreSinger ***@***.***> wrote: @KA7E <https://github.com/KA7E> Thank you! I have raised a PR to undo this change so I can run the CI and see if it gets resolved: #2670 <#2670> I do not think it will be necessary to merge this reverting PR, but it will just verify if this is causing the issue. Perhaps this change makes the ISPD testcases a bit harder, and we may just need to move the ISPD tests into their own nightly_test. — Reply to this email directly, view it on GitHub <#2668 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AM6ZD4UFKQMNBWTJXH5XMW3ZO6IOVAVCNFSM6AAAAABLUTSW4CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENJYGM2TONRYHE> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[vaughnbetz]

Thanks. Reverting while investigating makes sense.
If necessary, we can make a different version of the ispd arch (ispd_flexible or some such) for the legalizer tests.