{"id":700,"date":"2016-10-13T20:21:19","date_gmt":"2016-10-13T20:21:19","guid":{"rendered":"https:\/\/aciref.org\/?p=700"},"modified":"2016-10-13T20:21:19","modified_gmt":"2016-10-13T20:21:19","slug":"on-building-a-fast-r-environment","status":"publish","type":"post","link":"https:\/\/aciref.chpc.utah.edu\/?p=700","title":{"rendered":"On Building a fast R environment"},"content":{"rendered":"

\t\t\t\tIn this article we will describe how to build an R distribution from source
\nusing the Intel Compiler Suite.<\/p>\n

Why?<\/h4>\n

The execution of R code can be significantly sped up when it is built
\nwith a decent compiler and relies on fast mathematical libraries (mainly linear algebra).<\/p>\n

Requirements<\/h4>\n

The installation (vide infra) was performed on a machine running on the Centos7 OS.
\nWe also used the Intel Compiler Suite (Intel Parallel Studio XE 2017) and
\nits Math Kernel Library (MKL).<\/p>\n

Installation Process<\/h4>\n

Downloading the source code<\/h5>\n

cd \/uufs\/chpc.utah.edu\/sys\/srcdir\/R
\nwget http:\/\/cran.stat.ucla.edu\/src\/base\/R-3\/R-3.3.1.tar.gz
\ntar -zxvf R-3.3.1.tar.gz
\nmv R-3.3.1 3.3.1
\ncd 3.3.1<\/p>\n

Setting up the Compiler Environment<\/h5>\n

module load intel\/2017.0.098
\necho $MKLROOT
\nexport CC=icc
\nexport CXX=icpc
\nexport F77=ifort
\nexport FC=ifort
\nexport AR=xiar
\nexport LD=xild
\nexport CFLAGS=” -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC ”
\nexport CXXFLAGS=” -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC ”
\nexport FFLAGS=” -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC ”
\nexport FCFLAGS=” -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC ”
\nexport LIBDIR=\\
\n\/uufs\/chpc.utah.edu\/sys\/installdir\/intel\/compilers_and_libraries_2017.0.098\/linux\/compiler\/lib\/intel64_lin
\nexport LDFLAGS=” -Wl,-rpath=$MKLROOT\/lib\/intel64_lin \\
\n -L$MKLROOT\/lib\/intel64_lin -lmkl_intel_lp64 \\
\n -lmkl_core -lmkl_intel_thread \\
\n -Wl,-rpath=$LIBDIR -L$LIBDIR -liomp5 -lpthread ”<\/p>\n

Generation of the Makefile<\/h5>\n

cd \/uufs\/chpc.utah.edu\/sys\/builddir\/R; mkdir 3.3.1i; cd 3.3.1i<\/p>\n

# Configure (with BLAS & LAPACK MKL Support)
\n..\/..\/..\/srcdir\/R\/3.3.1\/configure –prefix=\/uufs\/chpc.utah.edu\/sys\/installdir\/R\/3.3.1i \\
\n –enable-R-profiling –enable-R-shlib –enable-memory-profiling \\
\n –enable-java –enable-shared=yes –with-blas=”$LDFLAGS” \\
\n –with-readline –with-tcltk –with-libpng –with-jpeglib –with-libtiff \\
\n –with-ICU –with-pic –with-x –with-lapack<\/p>\n

Output of the configure command<\/h5>\n


\n# ———————————————————————
\nR is now configured for x86_64-pc-linux-gnu<\/p>\n

Source directory: ..\/..\/..\/srcdir\/R\/3.3.1
\n Installation directory: \/uufs\/chpc.utah.edu\/sys\/installdir\/R\/3.3.1i<\/p>\n

C compiler: icc -std=gnu99 -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC
\n Fortran 77 compiler: ifort -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC<\/p>\n

C++ compiler: icpc -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC
\n C++11 compiler: icpc -std=c++11 -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC
\n Fortran 90\/95 compiler: ifort -axCORE-AVX2,AVX,SSE4.2 -O3 -ipo -qopenmp -fPIC
\n Obj-C compiler: gcc -g -O2 -fobjc-exceptions<\/p>\n

Interfaces supported: X11, tcltk
\n External libraries: readline, BLAS(MKL), LAPACK(in blas), curl
\n Additional capabilities: PNG, JPEG, TIFF, NLS, cairo, ICU
\n Options enabled: shared R library, R profiling, memory profiling<\/p>\n

Capabilities skipped:
\n Options not enabled: shared BLAS<\/p>\n

Recommended packages: yes
\n# ———————————————————————
\n<\/i><\/p>\n

Building the executable\/installation<\/h5>\n

make -j 16
\nmake check
\nmake install
\nmake pdf
\nmake install-pdf\t\t<\/p>\n","protected":false},"excerpt":{"rendered":"

In this article we will describe how to build an R distribution from source using the Intel Compiler Suite. Why? The execution of R code can be significantly sped up when it is built with a decent compiler and relies on fast mathematical libraries (mainly linear algebra). Requirements The installation (vide infra) was performed on […]<\/p>\n","protected":false},"author":3,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[],"class_list":["post-700","post","type-post","status-publish","format-standard","hentry","category-experience"],"_links":{"self":[{"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=\/wp\/v2\/posts\/700","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=700"}],"version-history":[{"count":0,"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=\/wp\/v2\/posts\/700\/revisions"}],"wp:attachment":[{"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=700"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=700"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aciref.chpc.utah.edu\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=700"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}