Fortran Code Porting problem

[postaquestion]

I have purchased DYNA3D fortran source code a few years back and this year purchased your fortran compiler for Mac OSX which I am very impressed by. Basically with only a few very simple modifications I am able to compile and run DYNA3D (which was supplied for installation on legacy Unix platforms e.g. DEC,SUN,IBM,SGI,HP,VAX,CRAY, not Linux or OSX), but I only have software to view one type of the binary output files... and unfortunately when requesting this binary output file DYNA3D crashes so I am not confident I have a valid baseline installation to work from.

DYNA3D was written by LLNL and sold to me by the US DOE with only very basic support, they have not even bothered to respond to my enquiry about this issue. I really need expert advice.

I would like to know which support department in Intel to contact regarding a quick once-over. The source code is well structured and easy to read, using consistent comment practices, but as it is legacy-code based it does have some ancient aspects, such as file management, it is however the more recent file format option provided by the code I am trying to enable. The problem I am encountering is simple as the rest of the code seems to work fine, but I am not a programmer and although have spent some time debugging the code and can see issues (bad memory access - error code 174) I do not have the experience to home in on the real issue.

I would be grateful if you could point me to the correct department to handle this issue. I am not sure if this is covered by the support provided by purchasing Intel Fortran last month.

I look forward to hearing the options on support on this code-porting problem I have.

[Steven_L_Intel1]

There really isn't an appropriate part of Intel to handle a request to debug a customer application. In the past we did have a division that offered paid consulting services, but it no longer exists. My recommendation is to use this thread to ask questions to help you learn to find the problem yourself.

The first step is to find out exactly where in the program the access violation is occurring. This will be in the traceback, or if you are running from the debugger, it should stop on the offending line. It is useful to know if the error is occurring at the start of a called routine as compared to inside a routine. Then, are the variables mentioned in the statement routine arguments? If so, what are the corresponding "actual arguments" passed to the routine?

You may find my article on access violations instructive. In particular, note the part about passing a constant to a routine which then tries to assign to the corresponding argument.

[Steven_L_Intel1]

By the way, are you running on a Mac? If so, I'll move this thread to the Linux/Mac forum. I ask because 174 is not an error code one would see on Windows.

[TimP]

I haven't looked at the NASA-furnished source code of DYNA3D in the last 20 years, and I doubt you will find anyone who has current knowledge of it. As you must be aware, several fully supported commercial applications are derived from it (LS-DYNA, Simulia/Abaqus, Pam-Crash, Radioss). Those have the benefit of 10's or more man-years of additional development. As far as I know, all organizations which perform such analyses have adopted one of these supported applications, regardless of their level of expertise.

As you haven'teven revealed your compiler options, I would suggest -i8 -r8 -O -g -assume fp-model precise,byterecl,buffered_io and be prepared to learn a lot more about debugging, if you are tackling this project for educational reasons. If your intent is to accomplishmechanical engineering work, you license one of those supported applications and take their training, which is available at competitive prices.

Even if you were an expert programmer, you wouldn't have time left to use the code, after you tackle building it.

If you are researching the use ofsuch analysisby NASA, here is a useful web search string: "LS-DYNA Challenger investigation."

[funkypuck]

Hi Steve, yes I'm on an Intel Mac and trying to use it for everything hence this porting effort (and now of force as my old windows box is a brick). I am very grateful to your quick replies and those of Tim18. Just to give an idea of the intent of this port, I am interested in developing the code to output new file formats compatible with CGI/rendering software - one reason for the Mac bias - and also later to develop a simple AI link into the code (to allow sensor analysis and actuator response). Apols for this appearing in the wrong forum, it's my first enquiry/post and was posted on my behalf by isn.support. Will attempt due diligence from here on in.