Lapacke_Openblas.md

Working with Openblas And Lapack on C

Usefull Resources and Prerequisite packages :

• Official OpenBlas Repository
• Official Lapacke Docuementation
• Example Repository for Openblas and Lapacke
• Installation either manually or with packages of the libraries is mandatory :
  pacman -S lapack lapacke lapack-doc <openblas or blas> cblas

Openblas

In order to use correctly the Openblas functions with C the cblas interface is mandatory.
Effectively the use of #include <cblas.h> and compiling with the -lcblas is enough to succesfully
run blas calls.
In order to tune the the Openblas library use :

export OPENBLAS_NUM_THREADS=threads       openblas_set_num_threads(threads);
export GOTO_NUM_THREADS=threads           goto_set_num_threads(threads);
export OMP_NUM_THREADS=threads            

List of functions:

1. cblas_zdotc_sub: dot product of two complex vectors
2. cblas_dgemm: matrix multiplication
3. cblas_dscal: scale a vector/array
4. cblas_dswap: Exchanges the elements of two vectors (double precision)
5. cblas_dcopy: Copies a vector to another vector (double-precision).
6. cblas_daxpy: Computes a vector-scalar product and adds the result to a vector.
7. cblas_dgemv: Multiplies a matrix by a vector (double precision).
8. cblas_dger: Multiplies vector X by the transform of vector Y, then adds matrix A (double precision).
9. cblas_ddot: Computes the dot product of two vectors (double-precision).

Lapack

Similarly with the openblas execution the inclusion of #include <lapacke.h> and linking with
-llapacke during compilation is sufficient to run lapack calls.

Some random Extra resources :

• Amd:: Developer Guides
• Agner:: Software Optimization Guides