Using full capcity of quadcore processor for calculating

[xk2379_52]

I try to use the MKL in an C programm compiled with Visual C++ 2005 on an quadcore pc.

At the moment the task is only using one of the four processors.

How can I used all four processors for my programm ?

[Tabrez_Ali]

Which MKL routines are you calling ?

The BLAS 3 routines are multithreaded and anything that depends on it should parallelize. Set the OMP_NUM_THREADS variable to 2 or more and try running your code.

You may or may not see speedup depending on the type/size of your problem.

[xk2379_52]

I using the LAPACK routines zheevd, zheev and dsyev for matrix diagonalization. Setting of OMP_NUM_THREADS don'nt show any effect.

[Michael_C_Intel4]

Which MKL version do you use? What is the size of the problem?

dsyev/zheev/zheevd have been threadedin MKL 10.0 quite good. There was a lack of parallelism before that.

[xk2379_52]

I use MKL 10.0 to diagonalyze a lot of 81 x 81 complex matrices.