search

RTimothyEdwards / netgen

Netgen complete LVS tool for comparing SPICE or verilog netlists
Other
110 stars 25 forks source link

Improve flattening #45

Closed d-m-bailey closed 2 years ago

d-m-bailey commented 2 years ago

@lankasaicharan Any idea why the CI check is failing?

ghost commented 2 years ago

@lankasaicharan Any idea why the CI check is failing?

@d-m-bailey its the 5th testcase (https://github.com/lankasaicharan/vezzal/tree/master/testcases/netgen/case5) which is failing. I am attaching the log file of it. vezzal_fail.log

This started failing when FlattenUnmatched() was edited. https://github.com/RTimothyEdwards/netgen/commit/de18ae85a4f0a96271a301a4704d42e7ac132963

d-m-bailey commented 2 years ago

@lankasaicharan Ok, so it's not necessarily a failure caused by my changes, right?

The netB.spice looks like it's missing devices inside subcircuits.

Contents of circuit 2:  Circuit: 'sky130_fd_sc_hvl__inv_8'
Circuit sky130_fd_sc_hvl__inv_8 contains 0 device instances.
Circuit contains 0 nets.

The current version of netgen does not flatten a circuit if the corresponding circuit does not contain any devices.

ghost commented 2 years ago

Right.

@RTimothyEdwards Is this something you are looking at or do you think I need to edit something in the testcase?

RTimothyEdwards commented 2 years ago

I'll check this today. The original intent (sometime way back) was that the "-blackbox" option would control the behavior with respect to empty subcells. I don't think it is workable to hard-code it one way or the other. But I think that right now, the "-blackbox" option, while parsed, doesn't do anything meaningful.

RTimothyEdwards commented 2 years ago

I ended up merging this one manually, but it's been done.