Abaqus 2019 + Visual Studio 2019 + OneAPI HPC

[crushing]

1. Installed Abaqus 2019

2. Installed Visual Studio 2019 community (selected the Desktop C++ development option)

3. Installed Visual Studio 2019 buildtools (selected the Desktop C++ development options)

4. Installed OneAPI 2021 Base kit

5. Installed OneAPI 2021 HPC kit

6. Ran the OneAPI 2021 command line for intel64 for Visual Studio 2019

7. Edited the abaqus command to call vars.bat with the arguments intel64 and vs2019

Abaqus verify fails with this in the fortran_make log:

Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.2.0 Build 20210228_000000
Copyright (C) 1985-2021 Intel Corporation. All rights reserved.

Microsoft (R) C/C++ Optimizing Compiler Version 19.29.30037 for x64
Copyright (C) Microsoft Corporation. All rights reserved.

main_13972.C
main_13972.obj : error LNK2019: unresolved external symbol abqmain referenced in function main
C:\Temp\verify\fortran_make.exe : fatal error LNK1120: 1 unresolved externals
Abaqus JOB fortran_make
Begin Compiling User Post-Processing Program
5/27/2021 12:43:04 PM
End Compiling User Post-Processing Program
5/27/2021 12:43:05 PM
Begin Compiling User Post-Processing Program
5/27/2021 12:43:05 PM
End Compiling User Post-Processing Program
5/27/2021 12:43:09 PM
Begin Linking User Post-Processing Program
5/27/2021 12:43:09 PM
Abaqus Error: Problem during linking - C:\Temp\verify\fortran_make.exe.
This error may be due to a mismatch in the Abaqus user subroutine arguments.
These arguments sometimes change from release to release, so user subroutines
used with a previous release of Abaqus may need to be adjusted.
The Abaqus Make execution procedure exited with errors

I've been at this for 2 days now. I have no idea what else I need to try. We had this working with previous versions (and various combinations of previous versions) of Abaqus, intel Fortran and Visual Studio, but now we can't get the licenses for those old versions and it seems like intel is going to make us upgrade to OneAPI, anyway. Has anyone made this combination work?

[Barbara_P_Intel]

There is a known bug introduced in the 2021.2 Fortran compiler regarding the case of the C routine names.  Could that be the problem?  Can't tell from the output you sent.

Here's the workaround. 

[crushing]

I'm not sure how to apply that workaround. I apologize for a lack of familiarity with Fortran: We've never had to dig this deep into this process to get Fortran to play nice with Abaqus or vice versa.

From what I can tell, this is the source code for the verification test from Dassault for Abaqus:

SUBROUTINE ABQMAIN
INCLUDE 'aba_param.inc'
WRITE(6,100)
100 FORMAT('PASS_Fortran')
RETURN
END

[Barbara_P_Intel]

The Fortran subroutine is fine.  Add this compiler option /names:lowercase to the link and you should be set.  If you use Visual Studio, follow the directions in the link I sent earlier to add it to the build.

I don't have Abaqus available to give you more direct instructions.  Others on this Forum do. I'm sure they'll chime in.

[JohnNichols · Accepted solution]

Run Subroutine in Abaqus2020 (step-by-step setup) - YouTube

Shows a great video on how to do it.  

Abacus is really straight out of the 1980s in terms of usability and friendliness. 

[JohnNichols]

https://www.youtube.com/watch?v=CQUMynkBezE

This shows how to do Fortran in the CAE -- it is also quite good. 

I find the people who makes these in India are slow and methodical and excellent as teachers. 

[crushing]

Barbara and John,

Thank you both so much: Things are running along smoothly now.

If it's okay, I would like to mark John's first reply with "Accept as solution" since that brought it all together,  but again, thank you both!

[JohnNichols]

No problem,  I really dislike abacus, but I am glad you got it running.  A program that makes this much money should have world class support and documentation.  

The input file is a tortured mess of 1970's style punch card input and could be trimmed down a lot. 

If I was on their board I would suggest that they look to the Fortran documentation as a way to do it better.  I still love the Fortran manuals - from MS that are printed and bound.  Sit beside my desk. 

Stay well and remember a FEM program is an approximation, it is not the answer.