Support for upcoming 55 nm pdk

[Andrea-Spitale]

Description

Hi, I've come across this (check slide 16) presentation held for fsic 2025, I think this is the same team responsible for iEDA.

Are you perhaps aware of this upcoming 55 nm open source PDK? Looks like they should release it this month. I am just curious to know if openroad would be open to support new PDKs.

Thanks!

Suggested Solution

No response

Additional Context

No response

[gadfort]

Where is it being released, the slide just says it will be released, but there is no link to where it may be found when they do release it.

[Andrea-Spitale]

You're right. I have sent some emails to both the iEDA team, which seems to have already done some runs with that PDK, and the icsprout Company. I'll post here once I have updates.

[maliberty]

We would be willing to support it if it is open. I've seen the same presentation a few times now (also at DAC) but not yet the PDK.

[QuantamHD]

The feeling I get is that it's "open" not "open source". All the presentations mention that it will be available in ECOS which is their web based iEDA environment.

I strongly suspect that's the only place it'll be. Rather than being released for standard consumption

[Andrea-Spitale]

So I got no response from both sprout and the iEda team. The iEDA studio platform doesn't seem to exist yet or it is not accessible from europe. If you prefer to close this thread until they share news, please do so..

[Andrea-Spitale]

Hi OpenRoad team, the pdk has been realeased! https://github.com/openecos-projects/icsprout55-pdk

What do you think?

@maliberty @gadfort tagging you cause I cannot reopen the issue and I hope this doesn't go unnoticed.

[gadfort]

Just a couple of notes from taking a look:

1. GDSs are missing so streamout is not possible.
2. There is no information about parasitics

[maliberty]

The do give the basic LEF parsitic info in https://github.com/openecos-projects/icsprout55-pdk/blob/main/prtech/techLEF/N551P6M_ieda.lef, ie

RESISTANCE RPERSQ CAPACITANCE CPERSQDIST EDGECAPACITANCE

(idk why they have two different tech lefs)

[maliberty]

https://github.com/openecos-projects/icsprout55-pdk/issues/2

[Andrea-Spitale]

Thanks for having a look. I'll be monitoring this issue closely, but unfortunately I cannot contribute as my knowledge is limited. I simply hope there will be support for it if the pdk team can provide you with enough information and files.

Thank you very much.

[gadfort]

OpenROAD generates an error on this line: https://github.com/openecos-projects/icsprout55-pdk/blob/493908790d18b702fd50f3015d0b9a72b3465464/prtech/techLEF/N551P6M.lef#L651

[povik]

I've done the bare minimum to set up synthesis in branch https://github.com/povik/OpenROAD-flow-scripts/tree/ics55-synthesis

[mi-and-core]

@gadfort I'm not involved, but it looks like "NONDEFAULTRULE virtuosoDefaultTaper" is a leftover from a more complete PDK. (POLY—like ACT, PP, NP, and NW(1)—is a MASTERSLICE in the currently released PDK.) A quick test shows that removing lines 644-652 from N551P6M.lef is sufficient.

Regarding N551P6M.lef / N551P6M_ieda.lef: *ieda.lef contains some additional information (e.g., CAPACITANCE and EDGECAPACITANCE in lines 72/73) and, for reasons unknown to me, occasionally different OFFSETS than N551P6M.lef (e.g., line 67).

[mi-and-core]

@povik Nice work. When you proceed to the next step of floorplanning, please temporarily remove lines 644-652 in N551P6M.lef.

[povik]

@mi-and-core at the moment I'm not pursuing setting up this PDK further.