-
Notifications
You must be signed in to change notification settings - Fork 9
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
Tests fail on MacOS aarch64 #64
Comments
Oh, that is not just insufficiently accurate, but rather completely wrong. I'm afraid I have no idea what could be causing this. |
Update 1Disabling OpenMP in the build process of finufft for MacOS with aarch64 resolves this issue. Update 2On my M3 Pro (12 Cores), the issue occurs only if OpenMP uses 11 or more threads. With Further, I have noticed one possible mitigation approach: Building finufft from Yggdrasil for Mac OS with aarch64 using clang ≤ v14 resolves the issue also for twelve OMP threads. However, the older clang produces a slower library... |
This PR fixes wrong test results in FINUFFT.jl on MacOS (see ludvigak/FINUFFT.jl#64)
Will be closed after JuliaPackaging/Yggdrasil#9423 has been merged. |
This PR fixes wrong test results in FINUFFT.jl on MacOS (see ludvigak/FINUFFT.jl#64)
With the finufft library in finufft_jll (v2.2.0) and copies build using BinaryBuilder from the current master https://github.com/flatironinstitute/finufft, running the tests fails on MacOS Sonoma 14.5 with an Apple M3 Pro (at least) due to insufficiently accurate results:
Collapsed error log
Building the library locally from source and linking to the FFTW and OpenMP libraries used by julia according to
restores the accuracy. There is no difference in the behavior whether FFTW or DUCC is used as fft-library.
The text was updated successfully, but these errors were encountered: