The problem for the substitute function of ippmEigenValuesVectorsSym_m_64f

Pham_Minh_N_ — Fri, 04 Mar 2016 09:22:19 GMT

Hi every one,

I would like to migrate some functions from IPP 7.0 of Intel Parallel Studio 2011 into Intel Parallel Studio XE 2016, but I have some problems about

ippmEigenValuesVectorsSym_m_64f function. According to the website below:

https://software.intel.com/en-us/articles/the-alternatives-for-intel-ipp-legacy-small-matrices-domain

ippmEigenValuesVectorsSym_m_64f can be replaced by LAPACK of Intel MKL Domain. I found the information of substitute function as link:

https://software.intel.com/en-us/node/469176

The LAPACKE_dsyev is the suitable function for replacing ippmEigenValuesVectorsSym_m_64f function. But when I take the example to verify, the result between old and new function are different.

Please see the source code and the result that are attached below

The result for Eigen Vector and Eigen Value of new function seem to be inversed.

Please give me some suggestions to solve this problem

Thanks,

Pham Minh Nhan

Hi Pham Minh Nhan,

Shaojuan_Z_Intel — Mon, 14 Mar 2016 21:47:32 GMT

Hi Pham Minh Nhan,

From your attachment, the two functions return the same results, but with different order: one in increasing order for values of Eigen Value, and the other in reverse order. For each eigen value, the Eigen Vector is shown in corresponding column. So the two functions are the same. Please let us know if you have any further questions. Thanks!

topic The problem for the substitute function of ippmEigenValuesVectorsSym_m_64f in Intel® Integrated Performance Primitives

The problem for the substitute function of ippmEigenValuesVectorsSym_m_64f

Hi Pham Minh Nhan,