Using Intel MKL with R (Similar to MATLAB Implementation)

[Maraki]

Hello,

I've been using the Intel MKL library in MATLAB successfully to optimize a time-consuming subfunction. To do this, I set up the compiler and flags in MATLAB using the following commands:

mex -setup
C:\Program Files (x86)\IntelSWTools\compilers_and_libraries_2020\windows

mex -v -g example.c -I"C:\Program Files (x86)\IntelSWTools\compilers_and_libraries_2020\windows\mkl\include" -L"C:\Program Files (x86)\IntelSWTools\compilers_and_libraries_2020\windows\mkl\lib\intel64" -lmkl_intel_lp64 -lmkl_core -lmkl_sequential -lmkl_cd

Now, I'm looking to achieve the same optimization in R, but I'm unsure of how to set up the Intel MKL library and compiler in R for a similar purpose. I've installed Intel oneAPI and  I've followed the following steps to set up my environment:

# Path to the Intel oneAPI toolkit 
oneapi_dir <- "C:/Program Files (x86)/Intel/oneAPI"
mkl_include_dir <- file.path(oneapi_dir, "mkl", "latest", "include")
mkl_lib_dir <- file.path(oneapi_dir, "mkl", "latest", "lib", "intel64")

# C code file
c_code_file <- "C:\\Users\\Desktop\\example.c"

#
output_executable <- "my_program"

# Define the compilation command with the output executable name
compile_command <- paste(
"icc -o", output_executable,
c_code_file,
paste("-I", mkl_include_dir, collapse = " "),
paste("-L", mkl_lib_dir, collapse = " "),
"-lmkl_intel_lp64 -lmkl_core -lmkl_sequential -lmkl_cdft_core -lmkl_intel_thread -qopenmp"
)

# Compile the C code
system(compile_command)

However, when I run this script in R I get this

system(compile_command)
[1] 127

The c-code is: 

#include <stdio.h>
#include <mkl.h>

int main() {
    // Define vector size and variables
    int n = 1000;
    double *x, *y;
    double result;

    // Allocate memory for the vectors
    x = (double*)malloc(n * sizeof(double));
    y = (double*)malloc(n * sizeof(double));

    // Initialize the vectors with some data (you can modify this)
    for (int i = 0; i < n; i++) {
        x[i] = i + 1.0;
        y[i] = 2.0 * (i + 1.0);
    }

    // Calculate the dot product of the vectors using MKL
    result = cblas_ddot(n, x, 1, y, 1);

    // Print the result
    printf("Dot product: %lf\n", result);

    // Free allocated memory
    free(x);
    free(y);

    return 0;
}

Could someone please guide me on the equivalent steps to perform this in R? I'm particularly interested in how to configure R to use the Intel compiler and link to the MKL library.

Thank you in advance for your help!

Best regards,

[ShanmukhS_Intel · Accepted solution]

Hi Maraki,

Thanks for posting in Intel communities. Thanks for elaborating on your issue.

We have tried running the shared code and below is the result. 

Dot product: 667667000.000000

Please refer to the below links for compiling R code with MKL, extended support, and other linking-related details.

How to use the BLAS and LAPACK libraries within Intel® oneAPI Math Kernel Library (oneMKL) to improve the performance of R. 

https://www.intel.com/content/www/us/en/developer/articles/technical/using-onemkl-with-r.html

Other important data which might be useful to you:

Building R with the Intel® Math Kernel Library (Intel® MKL) BLAS and LAPACK to improve the performance of those parts of R that rely on matrix computations. 

https://www.intel.com/content/www/us/en/developer/articles/technical/extending-r-with-intel-mkl.html

Linking Intel MKL to R

https://www.intel.com/content/www/us/en/developer/articles/technical/quick-linking-intel-mkl-blas-lapack-to-r.html

Could you please get back to us with the MKL version being used and your findings for the code being used by you? 

Best Regards,

Shanmukh.SS

[ShanmukhS_Intel]

Hi Maraki,

A gentle reminder:

Has the information provided helped? Could you please get back to us if you have any updates on your issue?

Best Regards,

Shanmukh.SS

[ShanmukhS_Intel]

Hi Maraki,

A gentle reminder:

Has the information provided helped? Could you please get back to us if you have any updates on your issue?

Best Regards,

Shanmukh.SS