-
Notifications
You must be signed in to change notification settings - Fork 181
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Running the Optcarrot benchmark iteratively in the same VM results in a core dump #919
Comments
The Unfortunately I cannot reproduce this on Fedora 24. All 50 runs work successfully and it ends with:
Which is just that extra I am running with the full path to bin/ruby: |
I see, the |
Update: My Notebook does not run in any problems either (though using TR 0.30.1).. strange. Its runs the same Ubuntu with a slightly different Kernel and CPU. Don't know how to solve this on the desktop, though |
I tried reproducing on a laptop and a server but couldn't on either. Could you try with:
It should dump some compiler graphs. You will see a message somewhere is the stdout. e.g.,:
If it doesn't reproduce with that command line, try removing the If you have a run that dumps graphs and reproduces, could you provide the |
Thanks for investigating, @gilles-duboscq! This is the cli-output when running the command you provided. Excluding
The files are attached to my comment: |
Thank you @Ichaelus ! The dump helped narrow down the part of the compiler that's doing the wrong transformation. I'm still looking into it. |
Hi @Ichaelus, I have a potential fix which i could test against a unit test which reproduces something similar to your issue but it would help if you could confirm that it really fixes your issue. To do so, could you replace |
I'm totally amazed! Can't stop grinning, since I tried to circumvent this error for a while now ;-) |
Thanks a lot for your help with diagnosing and testing this! I'll start the review process and hopefully the fix should be in the next GraalVM version. @eregon, i'm not sure what we should do with this issue here. Should it be closed since it's a Graal compiler issue and not a truffleruby issue? Or do you prefer to wait until there is a Graal version with the fix? |
@gilles-duboscq and @davleopo Thanks a lot for the help! I think it's best to wait until the fix is in Graal and then we can link it here, so we know the revision of the fix. |
Should be fixed in oracle/graal@6cfa502 |
Does this fix include additional code as the graal.jar you posted earlier on?
keeps delivering results, but the VM slows down enormously at ~ 220 calls. (x-axis: Seconds, y-axis: FPS) I know this is low-priority bug, but I felt like you could want to know about it. Michael |
The JAR i provided earlier was built by applying the same fix locally on the 0.30.2 branch. |
Ok. I'll try generate a 0.28 vs. 0.30.2 report in the next couple of days. |
@Ichaelus Could you open a separate bug for the slowdown? |
Ok we're at #959 now. |
Outline
Reproduction
ruby -S truffle-benchmark.rb
with (an rbenv version of) TruffleRuby 0.30.2.Error description
After ~ the 7th run, the JVM tries to dump its core. Exact message:
The referenced log file can be found here. Version 0.29 does not seem to be affected of this issue.
The text was updated successfully, but these errors were encountered: