C# ?gesvxx memory issue

[Gianluca_G_1]

Hello,
We are working in C# using MKL, just importing mkl_rt.1.dll and declaring functions in our code.
We use to work with huge linear systems with complex numbers and resolve this systems with functions like: LAPACKE_zgesvxx(..)

We have some issues when we try to resolve matrices that execede 2MB because the result is completly wrong!
(we made a simple test with a diagonal matrix with all (1,1j) in the diagonal and VTN of all 0.5 )

We saw there are two possible einterfaces: ILP64 interface uses the 64-bit integer type, while LP64 use the 32-bit integer. We know that the dll is x64 bit compiled but not sure about which interface is enabled.

We set the env variable MKL_INTERFACE_LAYER=ILP64.

Without effect.

is it possible to know if this variable is effective?

We are not C++ specialist so we would like to avoid working in C++ if possible.

Thank you for any advice you can give us.

BR
Gianluca

[ShanmukhS_Intel]

Hi,

Thank you for posting on Intel Communities.

Could you please help us with your environment details, so that we could assist you further with the required information wrt. mkl_rt.1.dll.Intel MKL uses lp64 for Intel-64 architechture.

In addition, please refer to the below link which might help you regarding the usage of LP64 vs ILP64 interfaces.

https://www.intel.com/content/www/us/en/develop/documentation/onemkl-windows-developer-guide/top/linking-your-application-with-onemkl/linking-in-detail/linking-with-interface-libraries/using-the-ilp64-interface-vs-lp64-interface.html

Best Regards,

Shanmukh.SS

[ShanmukhS_Intel]

Hi,

A gentle reminder:

Has the information provided helped? Could you please let us know if we could close this thread at our end.

Best Regards,

Shanmukh.SS

[ShanmukhS_Intel]

Hi,

We assume that your issue is resolved. If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.

Best Regards,

Shanmukh.SS