ifort: Setting-up on Fedora 32 Linux

[GS76]

Hi all,

I have a number of questions with regards my installation of parallel_studio_xe_2019_update1_cluster_edition on Fedora 32 Linux operating system.

Installation

First comment I know that Fedora is not a supported or tested flavour of Linux, but the compiler did install and run.

When I install it I get the following message:

No compatible Python* found
This system does not appear to have a Python* version 2.4, 2.6 or 2.7 installed. This may prevent operation of the GDB. For further details, please refer to the System Requirement section of the product Release Notes.

32-bit libraries not found
32-bit libraries not found on this system.
This product release requires the presence of 32-bit compatibility libraries when running on Intel(R) 64 architecture systems. One or more of these libraries could not be found:
libstdc++ (including libstdc++6)
glibc
libgcc
Without these libraries, the compiler will not function properly. Please refer to Release Notes for more information.

When I check if these items are installed they are and hence I am confused. I am new to Linux and Intel Fortran.

Running a program

I am able to compile a program running directly from within the installation path, as can be seen below:

The program:

program hello
print *, "Hello!"
end program

Logged into the root with sudo su:

The path shown below is: /opt/intel/compilers_and_libraries_2019.1.144/linux/bin/intel64

[root@localhost intel64]#./ifort /home/garystyger/Fortran/hello.f90 -o /home/garystyger/Fortran/hello.exe

[root@localhost intel64]# ./hello.exe

[root@localhost intel64]# Hello!

Environmental variables

So that is fine, however, I can not run the "ifort" from any other directory. Even after running the "psxevars.sh" script from the following directory: /opt/intel/parallel_studio_xe_2019.1.053.

I have tried running the "compilervars.sh" and "link_install.sh" scripts to no avail.

What am I missing or what have I not done right?

Any assistance would be much appreciated.

[mecej4]

The slip that is causing problems for you with the shell scripts is that you are running them by typing the name of the script at the command line instead of typing . or source followed a space and then the name.

Without that first element (the dot or 'source'), the environment that the shell script creates lasts only while the script is running. As soon as the script is done, the parent shell's environment is restored; that environment does not have the Intel tools along its PATH.

Please do not take the risk of attempting to use su or logging in as root to solve path and environment issues.

If you wish to generate 32-bit a.out-s, you will need to install the 32-bit GCC "compat" package, which can be downloaded using your system's package manager. A simple rule to follow: To ensure that your system has the 32-bit prerequisites, compile, link and run a simple C program with gcc -m32.

[GS76]

Hi mecej4,

Thank you for all your help.

However, I just want to be clear.
Are you saying I should do the following:
Be in the right directory (should this be run from root?):
source compilervars linux intel64

Would that be correct?

[mecej4 · Accepted solution]

The command can be run from any shell, with you logged in as a normal developer on that PC; not desirable: as root.

You can source the script from any directory. However, you will probably need to include the full path to the script if its directory is not already on the path. If, some time later, you close that shell and open another shell, you will need to enter the source command again. It is possible to modify your shell initialization to avoid having to do this, but that can create conflicts with the shells that you use when you wish to use software other than Parallel Studio.

Please read your release notes; it tells you how to access the compiler from a shell following installation of the package.

[Steve_Lionel]

Usually what you want is:

source /opt/intel/bin/compilervars.sh intel64

[GS76]

Thank you, Steve.

Much appreciated.