Also like what's the quantitative difference betwe...
# generalz
Zachary Curtis
08/13/2022, 4:49 PMAlso like what's the quantitative difference between the OpenROAD workflow and how ASIC design is done professionally?
a
Arman Avetisyan
08/13/2022, 5:42 PMOpenROAD is not as optimized as leading edge tools. No commercial support, your best bet is reading complicated source code. No support for many technologies. All bleeding edge technologies include PDK with ready to use support for different tools from different vendors, while OpenROAD is left mostly without out of box support
f
Farhad Modaresi
08/13/2022, 6:04 PMIs it possible for the designer to make OpenRoad more optimized for a particular design? If so, how could I do so
a
Arman Avetisyan
08/13/2022, 7:05 PMOpenLane has tools to run PPA analysis to pick the best strategy depending on your needs.
Also try using different stdlibs.
If you want to optimize the OpenLane in general then try adding multi vthreshold support. It should improve PPA.
Also, the routing and placement tools in openroad need some work too
z
Zachary Curtis
08/13/2022, 7:05 PMDo you have any recommendations on automated routing/placement?
a
Arman Avetisyan
08/13/2022, 9:09 PMI am not sure what you mean
m
Matt Liberty
08/13/2022, 10:30 PMProprietary tools have >10X the staff and are much older as well. I think the path forward is clear enough but the pace is limited by what the OSS community can muster. If you want to work on OR or OL development we can look for projects based on your skills/interests
z
Zachary Curtis
08/13/2022, 10:34 PMI'm interested, what sort of tools are needed?
Zachary Curtis
08/13/2022, 11:45 PMI'm generally interested if there is tooling that could help the OSS community.
m
Matt Liberty
08/14/2022, 12:54 AMI work on openroad so I can mostly speak for that. I know yosys and other tools could be improved as well. Do you have a software or eda background?
z
Zachary Curtis
08/14/2022, 12:57 AMNo eda, just hobby software. Like if there was a standalone tool that could be useful then I could tinker on it.
m
Matt Liberty
08/14/2022, 12:58 AM@proppy do you have any projects that might make sense?
p
proppy
08/14/2022, 2:51 AM@Matt Liberty I think some of the projects that we submitted for GSoC this year but didn't found any applicants are good candidates https://www.fossi-foundation.org/gsoc22-ideas#jupyter-notebook-add-ons-for-open-silicon-development https://www.fossi-foundation.org/gsoc22-ideas#python-bindings-for-openroad https://www.fossi-foundation.org/gsoc22-ideas#conda-packages-for-open-silicon-tooling
proppy
08/14/2022, 2:52 AMThere are also more ideas listed on the pages https://www.fossi-foundation.org/gsoc22-ideas
proppy
08/14/2022, 2:54 AM@Matt Liberty not sure if you have this for OpenROAD/ OpenLane yet, but it could be a good thing to have a dedicated label 🏷️ 'good-first-issue' / 'starter-project' for issues / feature request that are good I intro to the projects.
m
Matt Liberty
08/14/2022, 3:26 PM@proppy maybe you could turn these GSoC projects into issues and mark them as such
p
proppy
08/15/2022, 6:02 AM@Matt Liberty good idea, I believe we already tracking the packaging one in https://github.com/hdl/conda-eda, and we now have a repo to track the notebooks https://github.com/chipsalliance/silicon-notebooks
proppy
08/15/2022, 6:04 AMFor OpenROAD python, there is already https://github.com/The-OpenROAD-Project/OpenROAD/issues/1424 and https://github.com/The-OpenROAD-Project/OpenROAD/pull/1849 but I'm happy to split them with more granularity if it helps.
l
Lab Lecture
08/16/2022, 5:42 AMAs we know OpenRoad is well supported by DARPA $, will the volunteers who want to contribute get any $ ?
2 Views