-
-
Notifications
You must be signed in to change notification settings - Fork 5.5k
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
[LLVM] Upgrade to LLVM 13.0.0 #42602
Conversation
How are the TTFP implications? |
Without #42379 measurements of loading time on master of packages which load JLLs, like Plots.jl, are a bit meaningless |
Now merged, but on linux64 (sounds serious): line 499:
|
I don't think it is very useful to copy-paste CI logs for WIP PR into the comments unless you have something specific to say about it. I am sure the author of the PR knows how to find them. |
ecee36e
to
5b8d968
Compare
@pchintalapudi this looks similar to the bug we run into on #42572 |
That sounds like you accidentally initialized libjulia-internal twice, which I think we have seen the linker do because of the split with libjulia-codegen now |
Now just waiting on https://reviews.llvm.org/D114389, and getting LLVM (re-)packaged correctly in the build for Win32/Win64. |
Previously, this caused GlobalISel to emit invalid IR (a gpr32 to gpr64 copy) and fail during verification. While this shift is not defined (returns poison), it should not crash codegen, as it may appear inside dead code (for example, a select instruction), and it is legal IR input, as long as the value is unused. Discovered while trying to build Julia with LLVM v13: JuliaLang/julia#42602. Reviewed By: aemerson Differential Revision: https://reviews.llvm.org/D114389
Previously, this caused GlobalISel to emit invalid IR (a gpr32 to gpr64 copy) and fail during verification. While this shift is not defined (returns poison), it should not crash codegen, as it may appear inside dead code (for example, a select instruction), and it is legal IR input, as long as the value is unused. Discovered while trying to build Julia with LLVM v13: JuliaLang/julia#42602. Reviewed By: aemerson Differential Revision: https://reviews.llvm.org/D114389 (cherry picked from commit 18308e1)
aa04c69
to
bd990db
Compare
@nanosoldier |
@nanosoldier |
Your package evaluation job has completed - possible new issues were detected. A full report can be found here. |
@nanosoldier |
Something went wrong when running your job:
Unfortunately, the logs could not be uploaded. |
https://github.com/JuliaCI/NanosoldierReports/blob/master/benchmark/by_hash/15db64c_vs_c790938/report.md Looks okay. Sorry that this new AWS machine seems to be unreliable. Hopefully we can move back to dedicated hardware soon. @nanosoldier |
Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. |
Oh yeah that's extremely noisy :/ |
Non-scientific build benchmark: Before:
After:
|
Previously, this caused GlobalISel to emit invalid IR (a gpr32 to gpr64 copy) and fail during verification. While this shift is not defined (returns poison), it should not crash codegen, as it may appear inside dead code (for example, a select instruction), and it is legal IR input, as long as the value is unused. Discovered while trying to build Julia with LLVM v13: JuliaLang/julia#42602. Reviewed By: aemerson Differential Revision: https://reviews.llvm.org/D114389 (cherry picked from commit 18308e1)
llvm/llvm-project@llvmorg-13.0.0...JuliaLang:julia-13.0.0-3