Package Details: intel-mkl-bin 2019.5.281-1

Git Clone URL: https://aur.archlinux.org/intel-mkl-bin.git (read-only, click to copy)
Package Base: intel-mkl-bin
Description: Intel Math Kernel Library - shared libraries and headers only, as distributed by Anaconda
Upstream URL: https://software.intel.com/en-us/mkl
Licenses: custom
Conflicts: intel-mkl
Provides: intel-mkl
Submitter: chrisjbillington
Maintainer: chrisjbillington
Last Packager: chrisjbillington
Votes: 0
Popularity: 0.000000
First Submitted: 2019-11-18 03:27
Last Updated: 2019-12-09 14:38

Dependencies (0)

Required by (27)

Sources (2)

Latest Comments

chrisjbillington commented on 2019-12-10 14:26

@leomao, yes I'm aware. I don't think it really matters so long as there is no filename collisions with other packages, in which case moving things to /opt/ and adding an entry to ld.so.conf.d would allow them to coexist. I noticed that this package does duplicate some header files already in the fftw package, but they are in a subfolder of /usr/lib, so there is no collision at present (and the file contents are identical, so it doesn't matter which is used). But if a true collision comes up then I'll have to reconsider - on the other hand, conda packages will likely have the same issue, since they package a lot of the same libraries as might cause a conflict, and this package will just end up using their solution automatically.

leomao commented on 2019-12-10 08:11

@chrisjbillington, By the way, there are three different packages intel-mkl, intel-mkl-bin and intel-mkl-slim in the AUR. intel-mkl put the files in /opt/intel/mkl while the other two packages put them in /usr. This is a bit strange, but I am not sure which one is better...

chrisjbillington commented on 2019-12-09 14:41

@leomao, I haven't used the headers since I made this package to support numpy and scipy etc mkl builds. But I guess it's also a bit of a lie to say this package provides intel-mkl unless the includes are in it too! I have included them now just by pulling in the mkl-include conda package as you suggested.

leomao commented on 2019-12-09 09:53

To compile package with intel mkl, we need mkl header files from the mkl-include Anaconda package.