Apple silicon

finally Apple silicon Mac is available. Seems like it could be an ideal all-in-one solution that includes CPU + GPU + ML cores. It also explains why Apple has been so “strange” in the past several years dealing with Nvidia or AMD GPUs.

I wonder what are the difficulties in running Julia on Apple silicon? is there anything we could help?

Thanks.

It is a significant effort, which did get started a while ago. The new fortran compiler will need a lot of testing, and there’s the libunwind issue.

https://github.com/JuliaLang/julia/issues/36617

10 Likes

thanks for the effort!

M1 optimized TensorFlow is available! I wonder who have done that? Apple or TensorFlow community???

With TensorFlow being optimized on M1, does it imply that Python could also run on M1??

Really hope that we could have M1 optimized Julia!!!

1 Like

Python now runs on M1

It seems you can run Python per se, but

There is still no information about the availability of image processing and data science extensions of the Python core on the M1

as far as I understand, Fortran is a currently a big problem for R or Julia to run in M1.

does Python rely on Fortran?

I mean, if it does, that M1 Python package may include a M1 ready Fortran compiler? Maybe Julia could ride on it?

Python doesn’t require Fortran, but numpy and scipy do. AFAIK we’re all leaning on the same GCC fork right now.

2 Likes