Fortran77 open CFD software not executing 1

[ascocci]

Dear all,
I am trying to run a CFD analysis program written in fortran 77. The program is freely available at

https://www.dropbox.com/sh/8i0jyxzjb57q4j4/AABEMnM_zbfvRLcbpn2j8EkWa/MULTALL/multall program?dl=0

(Commall-open is needed in the same directory as multall-open to compile)
Further information on the software can be found at

https://www.dropbox.com/sh/8i0jyxzjb57q4j4/AABD9GQ1MUFwUm5hMWFylucva?dl=0

I use gfortran on windows 10 to compile the program and when I try to execute the program I encounter the error "This app can't run on your PC".
When trying to run the executable on linux (also compiled with gfortran) a segmentation fault occurs.
The command i use the compile the code is the following: gfortran multall-open-17.5.f -o multall-open
I can't find any way to execute the program after compiling it.
After many hours spent on trying to understand why the program does not work I found this forum and I hope you can help me. I don't know if the post on this part area of the forum is in the right place but i wouldn't know where to ask elsewhere. If you have any suggestion or know a better place to ask this question please let me know. If you have some time and want to try to run the program and let me know if you manage to i would be very thankful (only takes 3 mins).

thanks

Additional info on the program: when running the executable a simple file with written "N" should be present in the directory and a case test file should be passed to the program to run it (example: "multall-open.exe <case.dat") case files are available at

https://www.dropbox.com/sh/8i0jyxzjb57q4j4/AAC0ToS9B7xEKKDlhI0wpzLxa/MULTALL/multall-test-cases?dl=0

if comppiling multall-open-17.5.f a test case of the same version should be used.
This additional info is needed only once the program is running (would give an error message). I wasn't even able to run the program so if you can run the program (even without additional files and test cases) it would be a huge step for me!

 

[xwb]

Are you trying to run a 64-bit app on a 32-bit OS?

Do you have a mix of 32 and 64 bit DLLs?

 

[ascocci]

My os is 64-bit. The compiler I downloaded from

https://gcc.gnu.org/wiki/GFortranBinaries

(latest installer (dated 2014-06-29)) doesn't specify if it is for 32 or 64. It worked perfectly fine compiling other fortran codes on my os. About the DLLs i wouldn't know. How can I check? Thanks for the interest and trying to help me!

 

[xwb]

Do you have cygwin? If you do, type

file <dll>

Alternatively use the dependency walker. Run it on the exe and see which DLLs it picks up. If you are running a 64 bit program it will tell you which DLLs are 32 bit.

 

[ascocci]

Hi, thanks for the help.
I ran dependency walker and I got some strange results.
lots of errors "Error opening file. The system cannot find the file specified" for DLLS and also some modules with different CPU types were found.

This is a screenshot of the errors. I also put in the attachments the file that dependency walker produced.

 

[ascocci]

Strangely enough the other programs that I compile and that I am able to run have the same errors even if they seem to be running fine.

 

[mikrom]

...
I use gfortran on windows 10 to compile the program and when I try to execute the program I encounter the error "This app can't run on your PC".
...
When trying to run the executable on linux (also compiled with gfortran) a segmentation fault occurs.
...

I tried it on windows 7 and got similar error. But in Linux I was successful.

I used (in virtualbox) Linux Mint 17.3 64 bit with gfortran version 4.8.4 and compiled multall with this command:

gfortran -mcmodel=medium multall-open-17.5.f -o multall

Note, that without the switch -mcmodel=medium compilation failed with "relocation truncated to fit" errors

Then I called the program on one data-sample from multall-test-cases:

./multall < axial-cascade-q3d.17.1.dat > axial-cascade-q3d.17.1.dat.output

It produced some output in the output-file and creates other text-files during the computation - see the screenshot.
Unfortunately, I do not understand the numbers that the program has calculated

 

[ascocci]

Thanks Mikrom for letting me know you are able to run it! I really appreciate the effort for trying, this gives me hope! Is it possible for you to tell me with which settings you used to install Linux Mint? Did you have to allow for virtualization in the bios settings (I believe so for a 64 bit os)? Also when i tried to run Linux before in VirtualBox before it was super slow, do you have any suggestion on how to run it faster? Thanks again for the help!

 

[mikrom]

Hi ascocci,

I'm glad that I could help you a little bit, It was not so much effort. When I have a little time I'm rather solving practical problems than playing games.
As I wrote above, I used Linux running in virtualbox (VB) on windows. When you are on windows then download VB (

https://www.virtualbox.org/)

and install it. I have it on my workstation (W7) and on my laptop (W10) and it works nice. In VB you can run so many Linux virtual machines as you want and try them.

Then download a Linux ISO-image and install it in your VB - it's simple. There are tutorials on web (on youtube too) how to install Linux (Mint) on VB.
I'm using at home an older Version of Linux Mint 17.3 and therefore I have the same version in virtualbox on my Windows machines. But, I recommend you to use newer version which is now 18.3. The images are here:

https://www.linuxmint.com/download_all.php

I installed everything with default settings, nothing special. You can take Ubuntu too (

https://www.ubuntu.com/download/desktop),

because Mint is based on it.

So, then when you installed Linux Mint on VB, log into it and install gfortran. It's available from Software manager or Synaptic Package Manager, or when you open the terminal and try to run gfortran and it's not installed, then you will get a help. which tells you what you should to do: run a command
[pre]sudo apt-get gfortran[/pre]

Now about the bug with gfortran on windows:
This error has probably something to do with the security and UAC (User Access Control). We both (you and me) have only an older 32-bit versions of gfortran compiler installed. Maybe when you could find elsewhere on web a 64-bit version of gfortran for Windows, then it could work for you. Other option would be Cygwin:

https://www.cygwin.com/

... But IMHO the best of all is VirtualBox with Linux as I mentioned above.

 

[mikrom]

Did you have to allow for virtualization in the bios settings (I believe so for a 64 bit os)?

No, I have nothing changed in BIOS. I'm not allowed to do this because the PC and laptop are from the company I work for.

 

[mikrom]

Also when I tried to run Linux before in VirtualBox before it was super slow, do you have any suggestion on how to run it faster?

Maybe you should allocate more memory for your virtual machine (VM) in VirtualBox (VB). I think by default VB suggest you 2 GB. I have an workstation with 16 GB and Laptop with 8GB, so I increased the memory of VM from 2GB to 4GB. But you are right it's still slow.
I sugest:
1. solve your primary problem: i.e. you should be able to solve your CFD problem
2. then solve your secondary problem: i.e. try to increase the speed. IMO probably, you have then to install Linux natively.

 

[SchutzePhilip]

Hey guys,
I am also trying to get the Multall program to work. For me it also didn't work on Windows 10 (I also got the same error "This app cannot run on your PC"). Now I compiled it on linux but I do not get the test case to work. The program does not do anything when I try to open it, but when I run the test case like mikrom did it says "STOPPING BECAUSE FILE "intype" DOES NOT EXIST" inside the "axial-cascade-q3d.17.1.dat.output" file although it does create a file called "intype". Any help would be appreciated.
EDIT_1: I see there is information about this in the multall general description pdf under "RUNNING THE PROGRAM AND INPUT AND OUTPUT FILES". I will read through this first and retry.
EDIT_2: I followed the instruction in the PDF and got it to work.
Anyhow, thanks for thread - its useful to anyone who wants to try Multall.

 

[xwb]

Have you tried the "nothing to do with the program" fixes like

1) Disable smart screen
2) Enable sideload apps

It is something that W10 has to prevent attacks but sometimes it goes too far.

 

[mikrom]

Now I compiled it on linux but I do not get the test case to work.

Here is the website of Multall:

https://sites.google.com/site/multallopen/home

You can download the whole package here

https://www.dropbox.com/sh/8i0jyxzjb57q4j4/AABD9GQ1MUFwUm5hMWFylucva?dl=0

(click above right at the button Download and select Direct Download)
The test cases are in the subdirectory MULTALL/multall-test-cases or you can download them separatelly here:

https://www.dropbox.com/sh/8i0jyxzjb57q4j4/AAC0ToS9B7xEKKDlhI0wpzLxa/MULTALL/multall-test-cases?dl=0

I run the test case like mikrom did it says "STOPPING BECAUSE FILE "intype" DOES NOT EXIST"

You have self to create the file intype. It should contain only one character N (for new data format) or O (for old).
I used N. This is metioned elsewhere in a PDF-document which is part of the package.

 

[ascocci]

Now I compiled it on linux

Hi Philip, Which version of Linux are you using? Did you apply special settings? I still have to try to install Linux Mint as mikrom suggested but maybe you have some more insights too.
Also I was interested to know for which application you are using Multall as I haven't found anyone else on the internet that is trying to use this software. Are you available for a private dialogue? (Via email maybe?)
Let me know and thanks to all you guys trying to help!

 

[ascocci]

A few last questions to mikrom: when you ran the program did you get any error? and how much time did it had to run once it finished? was it printing anything on-screen in the terminal while it was running?

 

[mikrom]

Hi ascocci,
Have you already installed Linux?

Here in the file General-description.pdf on page 19 is short technical description how to run the program:

https://www.dropbox.com/sh/8i0jyxzj.../MULTALL?dl=0&preview=General-description.pdf

When I ran it I didn't get error. First when I ran It with only input file, i.e. like this:

./multall < axial-cascade-q3d.17.1.dat

I got huge output to the screen. Therefore in second ran, I redirected the output to the file, which I named axial-cascade-q3d.17.1.dat.output, i.e.:

./multall < axial-cascade-q3d.17.1.dat > axial-cascade-q3d.17.1.dat.output

But during the calculation the program created some additional files, some are binary and other text files with some data (see the screenshot above). Some of the files contain data for graphical visualisations (see PDF with description) (note, that the package contains folders with plotting programs for Linux and Windows).
From all test cases I first selected the smallest file, i.e. axial-cascade-q3d.17.1.dat. The calculation took only few minutes. I tried another test-case too, it was r37+samods-17.5.dat, but it took too much time. I had no patience, so after about 5 or 10 minutes I stopped the calculation.

 

[ascocci]

Hi mikorm, I managed to install linux mint and run the software! thank you for your suggestions and insights!

 

[DrMatt]

I was having the same problem (code not starting) when compiling multall on windows 10 with mingw 32-bit version.

Using the mingw64 gfortran I am able to get the multall code to run. However, I had to reduce the array size in the "common" file to be under the 2GB static memory allocation limit.

If you know of a version of multall which is reprogrammed with memory allocation instead of static common blocks, please share.

 

[mikrom]

Hi DrMatt,

Thank you for the advice about the MinGW 64 bit version. I have an older MinGW version which I copy from one Windows to another, so it's surely 32 bit. But the Linux I'm using is 64 bit, so the gfortran compiler on Linux is 64 bit too.

There are several vesions of multall available - 17.x and 18.x. I tried only 17.x.
Have you also tried the version 18.x, if it does not have the improvements you need?

Another possibility would be to ask the authors if there is a way to replace the COMMON blocks.

And finally, you have the source code, so you can try to build your own solution into the program.