Speed up RcppArmadillo: how to connect to OpenBlas in R package
I am working on an R package that uses RcppArmadillo. I am trying to use the faster matrix multiplication found in OpenBLAS. The documentation library C ++ armadillo says that if we have OpenBLAS in our car, the Armadillo will be used instead OpenBLAS BLAS. However, when I compile my R package, I get something like this:
g++ -m64 -std=c++11 -shared -L/usr/lib64/R/lib -Wl,-z,relro -specs=/usr/lib/rpm/redhat/redhat-hardened-ld -o PackageTest.so class1.o class2.o class3.o class4.o class5.o class6.o class7.o RcppExports.o class8.o class9.o class10.o -L/usr/lib64/R/lib -lRlapack -L/usr/lib64/R/lib -lRblas -lgfortran -lm -lquadmath -L/usr/lib64/R/lib -lR
Therefore, it compiles with parameters -lRlapack
and -lRblas
. How to change the files correctly Makevars
and Makevars.win
so that RcppArmadillo compiles the package with the option -lopenblas
? My attempt to solve this problem was to modify the file Makevars
like this:
PKG_LIBS = $(LAPACK_LIBS) $(BLAS_LIBS) $(FLIBS)
PKG_CXXFLAGS =-fopenmp -std=c++11 -lopenblas
PKG_CXX1XFLAGS = $(PKG_CXXFLAGS)
The package compiled with -lopenblas
, but the best way to do this?
source to share
This is an issue with the RedHat installation, which relied on the internal LAPACK sources for R when installing R --- plus the fact that RcppArmadillo uses everything that uses R.
It works differently on my Debian / Ubuntu based machine. Those. for
R> library(Rcpp)
R> cppFunction("arma::mat foo(arma::mat x) { return x + x;} ", depends="RcppArmadillo", verbose=TRUE)
I receive (inter alia)
g++ -Wl,-S -shared -L/usr/lib/R/lib -Wl,-Bsymbolic-functions \
-Wl,-z,relro -o sourceCpp_4.so file677111d81351.o \
-fopenmp -llapack -lblas -lgfortran -lm -lquadmath \
-L/usr/lib/R/lib -lR
and we see -llapack -lblas -lgfortran
as expected.
source to share