I am currently developping a user subroutine for ANSYS using ifort version 12.1.0 (The version needed to link the user subroutine to ANSYS) with the compiler installed on a Linux platform.
Toidentify the dependencies of the various subroutine and modules before compiling them, i used the command (All f90 files reside in the main folder when i do the operation.):
ifort -gen-dep:depend.mak -save-temps *.f90 1>>dump.std 2>>dump.err
ifort returns this into the depend,mak file for a few of the fortran file, mainly for subroutines which require a module:
statesType.o : \ statesType.f90 /usr/include/tensorstype.modintr /usr/include/tensorstype.modintr \ /usr/local/gcc-4.2.2/lib/gcc/x86_64-unknown-linux-gnu/4.2.2/include/tensorstype.modintr \ /usr/local/include/tensorstype.modintr \ /raid/home/public/intel/composer_xe_2011_sp1.7.256/compiler/include/tensorstype.modintr \ /raid/home/public/intel/composer_xe_2011_sp1.7.256/compiler/include/intel64/tensorstype.modintr \ /opt/intel/composer_xe_2011_sp1.7.256/mkl/include/tensorstype.modintr \ ./tensorstype.modintr tensorstype.modintr /usr/include/tensorstype.mod \ /usr/include/tensorstype.mod \ /usr/local/gcc-4.2.2/lib/gcc/x86_64-unknown-linux-gnu/4.2.2/include/tensorstype.mod \ /usr/local/include/tensorstype.mod \ /raid/home/public/intel/composer_xe_2011_sp1.7.256/compiler/include/tensorstype.mod \ /raid/home/public/intel/composer_xe_2011_sp1.7.256/compiler/include/intel64/tensorstype.mod \ /opt/intel/composer_xe_2011_sp1.7.256/mkl/include/tensorstype.mod \ ./tensorstype.mod tensorstype.mod
ifort seems to adequatetely identify the needed module but seems to add junk path before the file and additional .modintr files. Is there something i'm doing wrong ?
Please try a current compiler - the one you are using is five years old and we have fixed many bugs since then. If you can still reproduce the problem in a 17.0 compiler, let us know and include all source files necessary to show the problem.
I just checked ANSYS compiler requirements for my case and it says i can go up to INTEL 14.0.3. I will make the necessary arangements to update to this version and check what are the results.
Such statements simply mean that they (ANSYS) have tried that particular version. Later versions usually work equally well. However, since they have not tested those later versions, ANSYS do not make any claims about those. Try a current Intel compiler and, if it fails, you can go about obtaining a specific old version.
To read a recent thread of a similar nature, involving Matlab instead of Ansys, see https://software.intel.com/en-us/forums/intel-visual-fortran-compiler-for-windows/topic/684853 .