Package Details: r-mkl 3.6.1-1

Git Clone URL: https://aur.archlinux.org/r-mkl.git (read-only)
Package Base: r-mkl
Description: Language and environment for statistical computing and graphics, linked to the Intel(R) MKL.
Upstream URL: http://www.r-project.org/
Keywords: hpc mathematics modelling r statistics
Licenses: GPL
Conflicts: microsoft-r-open, r
Provides: r=3.6.1
Submitter: giniu
Maintainer: alexanderp
Last Packager: alexanderp
Votes: 19
Popularity: 0.55
First Submitted: 2010-05-06 00:10
Last Updated: 2019-07-11 20:06

Dependencies (26)

Required by (264)

Sources (4)

Latest Comments

1 2 3 4 5 6 ... Next › Last »

petronny commented on 2019-10-17 05:21

@xia0er R uses R-package/src/Makevars instead of the system Makevars when building packages. I find this when I try to build r-mxnet.

I can't recall exactly but I think there is a way to set the Makevars in userspace by editing something like ~/.R/Makevars.

Also someone filed an issue on arch4edu about that the prebuilt binary needs rebuild against the latest mkl. I'm working on it.

xia0er commented on 2019-10-16 20:09

My issue is fixed by editing /etc/R/Makeconf: FCLIBS_XTRA = -L/opt/intel/lib

Not sure why this is still needed, as I can see that /etc/ld.so.conf.d/intel-common-libs.conf has /opt/intel/lib included. I have also tried to set the environment variable export LD_LIBRARY_PATH=-L/opt/intel/lib before starting R, but it doesn't fix the issue.

Hope this helps others with similar issues.

alexanderp commented on 2019-10-15 21:58

@xia0er, looks like your system (container) might be broken. It's considered bad practice to manually manage software inside the container. Instead, recreate your base image.

xia0er commented on 2019-10-15 18:52

@alexanderp The docker image is based on archlinux/base and I update all packages to the latest version with pacman -Syyu before download and install r-mkl with this PKGBUILD (also tried @petronny's binary package) in my own Dockerfile. After building the image and run it in a container, r-mkl works fine as far as I can see, but fails when installing classInt (as a dependency for the package sf). I also tried installing the r packages in the Dockerfile, but it fails in the building process with the same error. Thanks for your help!

alexanderp commented on 2019-10-15 07:52

@xia0er, I guess the first question is how long ago the docker container was built and whether you updated software inside the container.

xia0er commented on 2019-10-15 05:24

Anyone experiencing issue when installing the classInt package on r-mkl? I am getting a strange error on an otherwise fully functional r-mkl installation:

... gcc -shared -L/usr/lib64/R/lib -Wl,-O1,--sort-common,--as-needed,-z,relro,-z,now -o classInt.so fish1.o init.o -lgfortran -limf -lm -lquadmath -L/usr/lib64/R/lib -lR /usr/sbin/ld: cannot find -limf collect2: error: ld returned 1 exit status make: *** [/usr/share/R//make/shlib.mk:6: classInt.so] Error 1 ERROR: compilation failed for package ‘classInt’ ...

But libimf.so is present in the /opt/intel/lib directory.

Besides, it also complains missing include dirs in a warning message, but, oddly, pointing to 2019.4.243

gfortran -fno-optimize-sibling-calls -fpic -m64 -I/opt/intel/compilers_and_libraries_2019.4.243/linux/mkl/include -c fish1.f -o fish1.o f951: Warning: Nonexistent include directory ‘/opt/intel/compilers_and_libraries_2019.4.243/linux/mkl/include’ [-Wmissing-include-dirs] fish1.f:100:72:

100 | DO 10 I=1,M | 1 Warning: Fortran 2018 deleted feature: Shared DO termination label 10 at (1) fish1.f:132:72:

132 | 40 IWORK(I,1)=1 | 1 Warning: Fortran 2018 deleted feature: DO termination statement which is not END DO or CONTINUE with label 40 at (1)

The classInt package installs successfully on a regular (non-mkl) R installation.

This happens when running r-mkl on an archlinux docker container, so there is no contamination from an earlier installation of mkl etc.

I appreciate it if anyone can help shed light on what may be happening here. Thanks!

The full stdout with error message can be found at https://gist.github.com/lmwang9527/48386437de1574aac45cb27725f129c7

alexanderp commented on 2019-07-26 07:45

Great @petronny. Thanks for your effort.

petronny commented on 2019-07-26 07:40

Hi, I provide a pre-built binary of this package at arch4edu.
Also please vote at this issue if you want more pre-built R packages.

petronny commented on 2019-07-24 07:06

Every time Intel MKL gets updated, compilation of R with icc needs to be adapted.

Actually I can deal with that, I can provide a well-built binary with every icc or r-mkl update. But I'll use gcc then since there is no gain when using icc.

alexanderp commented on 2019-07-24 06:40

Hi @petronny.

Every time Intel MKL gets updated, compilation of R with icc needs to be adapted.

I measured no gain when using icc, therefore this package defaults to gcc.

See some of my benchmarks here: https://github.com/alexisph/high_performance_r