This issue will need to grapple with some of the questions raised in chipsalliance/fpga-interchange-schema#12

Solution to chipsalliance/fpga-interchange-schema#12 will change requirements on this tool.

@litghost I am trying to get an initial working implementation for the FASM generator, initially tailored for 7-series, while trying to maintain it generic where possible.

One thing I stumbled upon is the generation of FASM features corresponding to route-thru pseudo PIPs. The problem is that the physical netlist does not have any information on whether a PIP is a pseudo one, (e.g. route-thru the OLOGIC site when there is no real BEL used). This would be relevant information to attach the correct FASM features when this kind of route-thru PIP is used and avoid outputting the route-thru PIP feature, given that it does not really exist. Would it make sense to add to the phys netlist schema an isPseudo field to the PIPs and emit that as well?

Would it make sense to add to the phys netlist schema an isPseudo field to the PIPs and emit that as well?

You should be able to consult the DeviceResource schema to figure out pseudo pips. There is still the problem of how to emit the FASM features for the pseudo pips, but for now I'd recommend doing a simple table lookup. Eventually the DeviceResource schema should contain enough information to emit the FASM features by understanding pseudo cell placements, but that is potentially hard, so could be considered out of scope.

Location in DeviceResource to determine if pip is pseudo pip:

https://github.com/SymbiFlow/fpga-interchange-schema/blob/4b3269148653b4e73c1c18c86d3c31d6d29e8453/interchange/DeviceResources.capnp#L223-L230

@litghost All right, I thought about getting information from the DeviceResources and the only initial problem I had was related to run-time, but I think I know what was wrong and I should be able to cut the overhead.

An initial version of the FASM generator is already merged and working for both xc7 and nexus. Further improvements and generalizations of the FASM generator can be discussed in follow-up issues. Closing