Significant Overhead if threaded MKL is called from OpenMP parallel region


my aim is to diagonalize quadratic matrices with different sizes dxd in parallel. To this end I wrote a for  loop. In each iteration the aligned memory (dependent on the dimension d) is allocated with mkl_malloc(). The matrix is filled and afterwards dsyev is called to determine the optimal workspace size. Then I allocate the (aligned) workspace needed with mkl_malloc(), call dsyev once again to diagonalize the matrices and deallocate the memory that was used for the workspace and to store the matrix (using mkl_free()). 

Using . instead of % as a component selector tool for a derived type


I accidently used the period operator (.) instead of % as a component selector tool in my Fortran code. This seemed to work perfectly fine with the Intel Parallel XE 2015 compiler. However, I could not find it documented anywhere as being a safe alternative for the % character. Is this a known feature or change to the standard?

I personally find it easier to read code with the period operator and would like to adopt this as a standard code-writing practice. Any suggestions or warnings will be welcomed.

PDT question

<p>I am trying some new things with Beta 2016 (W8.1, cygwin) and have several things working, including submodules, PDTs, and DTIO !

<p>I have encountered a couple of bugs and ICEs, which I have submitted, but this is a question.

Entwicklungstools abonnieren