-
Notifications
You must be signed in to change notification settings - Fork 76
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
Support Veclib of MacOS #350
Comments
the built in apple framework for BLAS/LAPACK is Accelerate, ndarray and ndarray-linalg are both capable of using it (you may have to add a simple build.rs file but it's not too bad once you know the magic words). here is the blas_src page which is a good starting point, it links to the apple accelerate page. source: not a maintainer, I use this library and have apple M1 laptop. |
Hello, Can you provide an example how to prepare a build.rs only for MacOS? I saw that both ndarray and ndarray-linalg rely directly on openblas-src, intel-mkl-src et.al. Thanks, Jianshu |
sure, here's a simple binary crate I used to test the expm routine I'm trying to get pulled in: basic M1 setup. Look at |
that was a personal helper binary, I figured it'd be easier to point you to it so you could pull out the |
Hello all, Please help if you can, this crate: https://github.com/jean-pierreBoth/annembed I tried your suggestion and use blas-src, accelerate as the backend, but it does not work, the 3 features, openblas-system is the only one that works on M1 (aarch63) platform. Can you please try at your end. The problem is with ndarray-linalg, ndarray can use accelerate as back end easily but I always have this error: error: linking with here I have CC using both the default clang, or gcc installed via brew, none of them work. Thanks, Jianshu |
Hello Team,
It seems openblas on MacOS can be replaced by the system default blas implementation called veclib, R language support it as the backend instead of default openblas et.al on MacOS systems. I am wondering whether this is possible because for new apple processors like M1 series, Intel-mkl is not supported and openblas is slower compare to default veclib (at lease in R experience it is). I cannot find so many info about veclib online so it may requires some additional investigation.
Thanks,
Jianshu
The text was updated successfully, but these errors were encountered: