I can observe the processor begin to execute the code from the boostrap ROM, which also on-die, but because it has no working memory, it promptly crashes. I would like to add that I ran full-chip GL simulation which did not have this problem.

Hi Tholin, did you preserve a complete set of your original OpenLane output artifacts and is the project open source? I'm looking for projects that aren't performing as intended to work on some timing visualisations to see if it was possible to hotspot design concerns using original flow output artifact data or to find out if the data indicates all is good (so maybe the data is wrong) This isn't something I'm due to deep dive into (a specific project, until tooling is in a useful state) for a few months I'm more in the canvasing period at this time as projects are coming back and testing is occuring.

I have no committed and RTL/artifact changes since tapeout

The default branch has commits 24d and 5m and 6m ago. Which commit id was the one used for fabrication? Did you download and save the full set of files generated by the flow separately to your own storage? Was the fabricated GDS taken from the output of a github action flow run? GHA artifacts are only stored by GH for around a month before they are destroyed.

As I said, the artifacts used for fabrication are the ones in the repo. They have not been touched since tapeout.

As I said, it is the flow OUTPUT artifacts that are needed, not the flow input sources (although it is expected they will be available if the project is open source). For example can you provide a link to the final SPEF file such as the one found at a path like

/runs/project_run_name/results/final/spef/top_level_macro_name.spef

which are in the OUTPUT artifacts from the OpenLane run used to generate the GDS that was used for submission.

And the ODB files used to create the SPEF? As I understand it the repo is a composition project used to glue macros together to produce the final GDS needed for fabrication. Each macro is an output result of a seperate OpenLane run, it is both the final outputs and all the working files I am after (all part of the output artifact fileset). Since if an issue is found it should be possible to follow back the flow intermediate data to find the cause ( which is the purpose of the exercise for me, not to fix a project but to improve things in reporting and detection for the next project )