update OR version and update for API changes #756
Conversation
Signed-off-by: Matt Liberty <[email protected]>
|
@donn I can't tell exactly what is failing here - there is tons of output. I don't know if these are real or not. Any hints? |
|
These are definitely real. What you should typically do is check a failing design's report at the end of each run:
Sometimes, the failures are because the designs are failing when compared to an earlier benchmark, but most likely failures are just when OpenLane fails outright. In which case, you need to download the logs. You can't download them per-design because that's just how GitHub works, so what you can do is download all of the run_tarballs.zip file from the CI run:
Inside the tarball for each run, there's a file called flow_summary.log. With the APU as an example, here's the error:
In this scenario, it appears that there have been some regression in OpenROAD causing designs that were previously working to fail. When updating OpenROAD, what I typically do is run |
|
Donn, the documentation you just wrote sounds like a great starting point
for a "Debugging CI" page in the developer documentation or wiki!
(Hopefully something like development docs exists somewhere?)
I would just start with copying your comment verbatim and then spend **no**
more then 10 minutes cleaning it up.
**Only** put more effort into improving the page when you share a reference
to it.
…On Sun, Dec 12, 2021, 5:30 AM Donn ***@***.***> wrote:
These are definitely real. What you should typically do is check a failing
design's report at the end of each run:
[image: IMG_0127]
<https://user-images.githubusercontent.com/12652988/145714140-f3ac5a57-c9d2-4cee-922e-a36eff87d3fa.jpeg>
Sometimes, the failures are because the designs are failing when compared
to an earlier benchmark, but most likely failures are just when OpenLane
fails outright. In which case, you need to download the logs. You can't
download them per-design because that's just how GitHub works, so what you
can do is download all of the run_tarballs.zip file from the CI run:
[image: IMG_0128]
<https://user-images.githubusercontent.com/12652988/145714203-a3b68128-b65a-4de4-8da2-7f68a4f44c05.jpeg>
Inside the tarball for each run, there's a file called flow_summary.log.
With the APU as an example, here's the error:
[image: IMG_0126]
<https://user-images.githubusercontent.com/12652988/145714091-638cf14c-c8ef-41c0-a85a-8f938d6597b9.png>
In this scenario, it appears that there have been some regression in
OpenROAD causing designs that were previously working to fail.
When updating OpenROAD, what I typically do is run make fastest_test_set
locally first as a sanity check, then see which designs fail, then package
tarballs and pass them on to you guys. I can do that tomorrow if you'd like.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#756 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAAFFXAI5XBQBRS4PES43NTUQSPXJANCNFSM5J3IEL5A>
.
Triage notifications on the go with GitHub Mobile for iOS
<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675>
or Android
<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.
|
|
Agreed, @mithro. I'll see what I can do there. APU issue is The-OpenROAD-Project/OpenROAD#1419, isolating the rest... |
|
{APU, wbqspiflash, zipdiv, xtea, digital_pll_sky130_fd_sc_hd, usb_cdc_core, usb} are all failing because of OpenROAD#1419 |
Signed-off-by: Matt Liberty <[email protected]>
Signed-off-by: Matt Liberty [email protected]