Eigen multi-thread operations

Question

Eigen is a C++ template library for linear algebra ( site ). According to the documentation, newer Eigen versions support the usage of OpenMP to take advantage of multi-threading characteristics. Their official documentation mentions that some operations can be ran in paralell ( documentation ).

However, which -concrete- operations are susceptible of these advantages is not mentioned in any place. The only place I have been able to find something states that matrix and vector multiplication can be ran in parallel, but no more information is provided ( here ).

I would appreciate if someone could share a bit of knowledge about this uncertainty.

Answer 1

From Eigen's source directory:

$ find Eigen -type f | xargs grep -i openmp

...shows that only the general matrix by matrix product (GEMM) is using OpenMP (for now!).