LAPACk and BLAS libraries Command line flags

[singh_balwinderd-p-l]

Hi All,

I am a newbie and would like to use LAPACK and BLAS libraries within my fortran code. I want to use command line instead of using Visual Studio for compiling and runnung my program. Would you guys please suggest me the best way to deal with this?

I went through the documentation and it was confusing for me. I was not able to figure out the flags I need to compile and run my program.

Thanks.

[TimP]

The link advisor in the right hand column of this forum should get you started.

[ArturGuzik]

Quoting - tim18

The link advisor in the right hand column of this forum should get you started.

Tim points you in the right direction. First start with link advisor. Link here. Then you can take a look at provided (together with examples) makefile(s). Inside you'll have a whole range of options/switches etc.

When building from command line please do not forget to run batch setting Build Environment for your compiler before building the app.

A.

[singh_balwinderd-p-l]

Thanks for your responses.

I went to the provided link, but I am not exactly sure about few fields there. Following were my selections:

Select OS: Windows
Select processor architecture: IA32
Select compiler: Intel or Intel Compatible

I think the next one ("Select dynamic or static linking") should be "dynamic" as I am mentioning the flags at the compile time instead of mentioning them within the program (Am I right?). For the next fields :

Select sequential or multi-threaded version of Intel MKL:
Select OpenMP library:
Select cluster library:
Select MPI library:

I dont have any idea. I am running a simple program which calculated eigenvalues from LAPACK's functionDSYEV.

Can you guys please help me with this?

Thanks.

[TimP]

On several of these choices, you could try it both ways, to see what you like.
If you don't have a reason to try static, dynamic link should be fine.
You would probably try the sequential version first, then you shouldn't get any questions about parallel libraries. Once you have it working, you would try the multi-threaded, in case it gives you speed-up.
In multi-threaded, if you have a choice of libiomp (compatibility with Microsoft OpenMP), that would be the normal choice.
I doubt you want to go down the cluster/MPI path.