Problem : I am having trouble installing my Intel Compiler Pro v11.1 on OpenSUSE


Environment : This article was written for OpenSUSE 11.2 and 11.1 but should apply to 11.0 installations as well.  This guide applies ONLY to the Intel Compilers Profession Version 11.1 and greater.  If you have an older Intel compiler, please upgrade to 11.1.  Compatibility of OpenSUSE 11.2 with anything other than Intel Compiler Pro 11.1 is not guaranteed to work - consider getting an older OpenSUSE 10.x or older version to work with your older Intel compiler.


Root Cause : As of this writing OpenSUSE 11.2 is not officially supported by the Intel Compiler Professional edition. The procedures herein are intended to assist the user configure OpenSUSE to support installation of the Intel Compiler Professional Edition.  However, this is not a supported configuration.  Please see your Release Notes document for a list of officially supported Linux distributions and versions.

The default OpenSUSE installation may not have all the packages needed to support software development (gcc, g++, and binutils packages along with header files).  Also, the Intel compiler released versions are targeted and built for distributions available 12-16 months before the compiler is actually released to customers.  Thus, by the time development is completed for a particular compiler version, many Linux distributions have progressed forward one or two versions.  Intel tries to stay current with Linux distributions, but it often takes us many months after a distribution release to add support.  See your Release Notes from the compiler documentation to determine supported distributions and versions.


Resolution : Add several packages to your OpenSUSE installation PRIOR to installation of the compiler.


Procedure:

00) Remove any previous failed installations:
If you have an installation of the Intel Compiler Pro v11.1 on your system from a failed attempt, uninstall it. 
/opt/intel/Compiler/11.1/046/bin/[ia32 | intel64]/uninstall_cpro[c | f].sh

substitute your version for "046" above if you have another version besides 11.1.046


0) First determine whether you have installed a 32bit or 64bit version of OpenSUSE.  Do this with the uname -a command:

uname -m

which will return:
x86_64   if you have a 64bit installation
i686  if you have a 32bit installation

Based on whether you have a 32bit or 64bit installation, choose one of the 2 procedures below.




1)  i686 32bit Installation

a) login as root or super user.
If you do not have the root password or priviledge, check if you have SUDO priviledges. 
If you do not have either root/admin or SUDO priviledges STOP.  You will be unable to install the Intel compilers on this installation until you are able to add system-wide packages as a root or admin user.
If you use SUDO to gain access, open a root window with this command:
sudo bash

b) Add Base Development and C/C++ Development
start Yast2 in interactive mode.  From the root window:   yast2
When the graphical admin tool Yast2 opens, there is a button (lower right) for "Software Manager".
Click Software Manager button to open the Software Manager tool. 
On this Software Manager window, find the "Filter" pull-down.  Set Filter to "Patterns" to display groups of packages in the left-hand list of package groups.  Scroll down in this list to find the group named "Development".
Under the "Development" group, select the following package sets to install:
"Base Development"
"C/C++ Development"

in addition, if you are installing the Intel C++ compiler and wish to use Eclipse integration, also install the "Integrated Development Environments"/"Eclipse" package set.

Once these package sets are select, select "Install" to install the developer packages on your system.

c) Install 32bit gcc and g++ 3.3 and make the default
From the terminal window, as root:

zypper install gcc33-32bit
zypper install libstdc++33-32bit

Try the installation of the Intel Compiler at this point.  If this does not work, consider doing the following:
rm /usr/bin/gcc
ln -s /usr/bin/gcc-3.3 /usr/bin/gcc

You may now proceed with an installation of the Intel Compilers.


2) x86_64 64bit OS Installation

a) login as root or super user.
If you do not have the root password or priviledge, check if you have SUDO priviledges. 
If you do not have either root/admin or SUDO priviledges STOP.  You will be unable to install the Intel compilers on this installation until you are able to add system-wide packages as a root or admin user.
If you use SUDO to gain access, open a root window with this command:
sudo bash

b) Add Base Development and C/C++ Development
start Yast2 in interactive mode.  From the root window:   yast2
When the graphical admin tool Yast2 opens, there is a button (lower right) for "Software Manager".
Click Software Manager button to open the Software Manager tool. 
On this Software Manager window, find the "Filter" pull-down.  Set Filter to "Patterns" to display groups of packages in the left-hand list of package groups.  Scroll down in this list to find the group named "Development".
Under the "Development" group, select the following package sets to install:
"Base Development"
"C/C++ Development"

in addition, if you are installing the Intel C++ compiler and wish to use Eclipse integration, also install the "Integrated Development Environments"/"Eclipse" package set.

Once these package sets are select, select "Install" to install the developer packages on your system.

c) Install gcc and g++ 3.3
From the terminal window, as root:

zypper install gcc33
zypper install libstdc++33
zypper install libstdc++33-32bit

At this point you are ready to install the 64bit versions of the Intel Compiler.  But before you do, read below:

d) Determine if you want or need to develop 32bit applications
Up to this point you have everything you need to install the 64bit Intel compilers and develop 64bit applications.  If this is all you need, locate the Intel compiler tarball named *_intel64.tgz, for example, l_cprof_p_11.1.046_intel64.tgz and install. 
IF you want to be able to create 32bit applications AND 64bit applications:

zypper install gcc33-32bit


You may now proceed with an installation of the Intel Compilers.  Use the combined IA-32 and Intel64 package, for example l_cprof_p_11.1.046.tgz.


Finishing up:  Usage

Once the compilers are installed, it is necessary to source the ifortvars.sh or ifortvars.csh file to properly set the user environment to use the Intel compilers.  This (these) commands can be placed in a user's ~/.bashrc or other system login file (fortran 11.1.046 example shown below, bash user):

64bit compiler selection:
source /opt/intel/Compiler/11.1/046/bin/ifortvars.sh intel64

32bit compiler selection:
source /opt/intel/Compiler/11.1/046/bin/ifortvars.sh  ia32

C/C++ users: substitute "iccvars" for "ifortvars" above.

Other versions: substitute your actual minor version for "046" above. 
Also, if you are a CSH or TCSH user, use ifortvars.csh or iccvars.csh as appropriate.


It is hoped this guide proves helpful for installing the Intel Compilers Professional Edition on an OpenSUSE installation.  Please add comments to this article if you have suggestions or tips to improve this guide.

Reference Number :  DPD200110877


Version : 11.0, 11.1 or Intel® Parallel Composer


Operating System : Windows*, Linux*, Mac OS X*


Problem Description : The OpenMP* 3.0 standard now supports using STL iterators for OpenMP loop bounds.  However, the Intel® C++ Compiler does not parallelize code like the following:

#include <vector>

void iterator_example()
{
  std::vector<double> vec(23);
  std::vector<double>::iterator it;

#pragma omp parallel for default(none) shared(vec) 
  for (it = vec.begin(); it < vec.end(); it++)
  {
    *it = 1.0;// do work with *it //
  }
}

The compiler will not give an indication (as it should) that the loop was parallelized for OpenMP*.  If you examine the code, you will see that the compiler generates a serial version of the loop.  This is because of an issue with the compiler using function calls on loop bounds that are inlined causing the compiler to not recognize the loop as being a validly formed loop for parallelization.


Resolution Status : This will be resolved in an upcoming compiler update.



[DISCLAIMER: The information on this web site is intended for hardware system manufacturers and software developers. Intel does not warrant the accuracy, completeness or utility of any information on this site. Intel may make changes to the information or the site at any time without notice. Intel makes no commitment to update the information at this site. ALL INFORMATION PROVIDED ON THIS WEBSITE IS PROVIDED "as is" without any express, implied, or statutory warranty of any kind including but not limited to warranties of merchantability, non-infringement of intellectual property, or fitness for any particular purpose. Independent companies manufacture the third-party products that are mentioned on this site. Intel is not responsible for the quality or performance of third-party products and makes no representation or warranty regarding such products. The third-party supplier remains solely responsible for the design, manufacture, sale and functionality of its products. Intel and the Intel logo are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. *Other names and brands may be claimed as the property of others.]