Prerequisites:

To build an application that uses the distributed memory feature of Coarray Fortran on Windows, you must have:

• Intel® Visual Fortran Composer XE 2011 Update 6 or newer
• A license for Intel® Cluster Toolkit, available separately or as part of Intel® Cluster Studio 2011

• A Windows-based cluster running Microsoft Windows HPC Server 2008* or Microsoft Windows Compute Cluster Server 2003*

Configuration Setup:

To run distributed applications you must have an established cluster, and the Intel® Fortran redistributable package must be installed on all target nodes in that cluster. Target nodes are cluster nodes that will be participating when your application is run.

You must create a file containing the names of all target nodes in the cluster; this is often called machines.Windows, but you can use any name.

Before running your application, you must register your Windows password with the Intel® MPI Library (part of Intel Cluster Toolkit). This needs to be done from an Intel Cluster Toolkit command window. It only needs to be done on the originating node, and only needs to be done once, until your password changes. The command will prompt you for your password:

mpiexec -register

Validating your configuration:

You must be able to successfully run an MPI job under Intel® MPI in order for a distributed coarray application to run. At the end of this document is an example "hello world" program. It is highly recommended that you successfully run this simple program on your cluster, across multiple nodes before attempting to run a distributed coarray application.

Build the hello_mpi.f90 program in an Intel Cluster Toolkit command window.

Run it with the command:

mpiexec –machinefile machines.Windows –n 4 hello_mpi.exe

Confirm that it ran as expected. The command line switch "-n 4" sets the number of images that are to be executed; you can use any number you want.

Build your application:

You are now ready to build your coarray Fortran application. The command is straightforward:

To run your program:

You need to pay attention to some details, or your program will only run on a single node.

The environment variable FOR_COARRAY_MACHINEFILE must be set to the full file path of your machines.Windows file. This only needs to be done on the node where you initially run your program.

The executable image of your program MUST be in a directory that is shared between all nodes in the Windows cluster, or the other nodes will not be able to run it. The full path of the executable, including the drive letter, must be valid on all cluster nodes.

Once these are set up, you simply have to run your program:

myprogram.exe

If there are command line inputs to your program, you can add them to the command line, such as:

myprogram.exe arg1 arg2

More things to think about:

Any input files to your program must be in a shared location so that all images on all nodes can access them.

It is recommended that all output files be generated into a shared location, but that may not be relevant to all programs.

Example Hello World program to test MPI

Program hello_mpi
   implicit none
   include "mpif.h"
   integer :: size, rank, ierr, len
   integer :: status
   call mpi_init(ierr)
   call mpi_comm_size(MPI_COMM_WORLD,size,ierr)
   call mpi_comm_rank(MPI_COMM_WORLD,rank,ierr)
   call MPI_GET_PROCESSOR_NAME(name, len,  ierr)
   write(6, "(*a,i3))") "MPI: size=", size,  "rank =", rank
   write(6,*) "host is", trim(name)
   call mpi_finalize(ierr)
   end program hello_mpi
   

Question and Answers:

Q: If I install Intel® Fortran Composer XE, will my old version still be there?

A: If by old version you mean 11.1, etc., then yes. The Visual Studio integration for Intel® Visual Fortran Composer XE 2011 gives you access to versions 11.0 and 11.1 through Tools > Options > Intel Visual Fortran > Compilers. Note that there are separate entries for 32-bit and 64-bit development.

On Linux* and Mac OS X*, the new installation will default to /opt/intel. Under this directory you will see your older 11.x compilers under the Compiler/11.1/xxx/ directories. The new Intel® Composer XE will have physical directory /opt/intel/composerxe-2011.12.1.122 (or similar): deciphering "12.1.122" is major version 12, Update 1, minor version 122. Along with this directory, you will see the new generic directory link "composerxe/" which links this generic path to the last version of the compiler installed on the system. These generic paths allow one to 'source /opt/intel/composerxe/bin/compilervars.sh intel64' and be assured that this path will exist and will link to the latest version compiler installed on the system. This provides a version-independent, default path to compilervars.sh for those who don't want to rewrite their makefiles or shell startup files for each new version installed on the system.



Q: I have been using Fortran for scientific applications since the 1980s, but am not a professional programmer. What is the most efficient way to learn how to properly use new features?
Q: Do you provide some level of guidance to write a parallel processing program?
Q: Is there a refresher course available to bring me up to speed from earlier versions of Fortran to the latest version?

A: Intel does not offer training on the Fortran language. There are several good books on the language which are more tutorial in nature. A list of books we have seen can be found in the compiler documentation under Introduction > Related Information



Q: Would it be possible to change the default N of images (Coarray Fortran) to 1, instead of the number of available threads?

A: Yes. This can be specified as an option (-coarray-num-images=1 on linux or /Qcoarray-num-images:1 for Windows) when compiling the main program or, before the program is run, by setting the environment variable FOR_COARRAY_NUM_IMAGES to be the number of images you want. This can be more or fewer than the number of available threads.



Q: Will there be extensions to Intel Fortran to permit programming GPUs using Fortran?

A: Intel will be extending its Fortran and C++ compilers to encompass Intel's future "Many Integrated Core" products. We do not have plans at this time to support GPUs from other vendors.



Q: What is the relationship between OpenMP and Coarrays?
Q: How did you implement coarrays? Does it use OpenMP?

A: We use Intel® MPI Library as the underlying support for coarrays, both for shared memory and distributed memory configurations. Intel MPI Library is a highly optimized implementation of MPI that supports a wide variety of interconnect types.

At present, we do not support the use of other parallel methods such as OpenMP or direct MPI calls in an application that uses coarrays. We recognize that customers may want to use multiple technologies and will be testing this in the future. You are welcome to try it yourself, but we don't promise it will work.



Q: Is it possible to have Intel Visual Fortran in Windows without Microsoft Visual Studio*?

A: Intel Visual Fortran relies on tools and libraries that come from Microsoft* Visual Studio*. If a system does not have a supported version of Visual Studio installed (2005, 2008 or 2010), we will install an environment based on Visual Studio 2008 Shell which supports Fortran development. You do not need to use the Visual Studio graphical development environment – you can still use the command line to build applications – but the only debugging environment supported is Visual Studio's.



Q: Is this compatible with the Intel® Cluster Studio XE?

A: Intel Cluster Studio XE 2011 includes Intel [Visual] Fortran Composer XE with full capabilities, including support for distributed memory coarrays.



Q: Can we expect improvements to the debugging experience when moving to VS2010 and Intel Visual Fortran Composer XE?

A: The debugging experience is improved continually, and we have resolved some known issues in Intel Visual Fortran Composer XE. A move to Visual Studio 2010, on its own, has no specific benefits here.



Q: Does Intel® Parallel Studio support Intel Visual Fortran?

A: Intel Parallel Studio is a C++ product, and its analysis tools do not support Fortran. Intel Parallel Studio XE includes both Intel C++ and Intel Fortran, as well as analysis tools that support both languages.



Q: I have found polymorphic entities are not very useful w/o sourced allocation. Any idea when it might come out.?

A: We can't say at this time when this feature will be added, but we are working on it.



Q: Is there any possibility of getting the documentation in pdf files?
A: Not at this time.



Q: Is Intel Visual Fortran compatible with Visual Studio 2005*?
A: Yes.



Q: Who besides Intel is involved in developing the Fortran standards?

A: You can view the membership of the US Fortran Standards committee at http://www.j3-fortran.org/. Vendors who are members of this committee are Cray, IBM, Intel and Oracle (Sun). A representative from NAG is a committee member, but not formally representing NAG. A member of the gfortran development team regularly attends meetings as well.



Q: Are there any plans to support Intellisense features in Fortran Composer XE?

A: We have had requests for some Intellisense features. It would help us if you would let us know, either in the user forum or an Intel® Premier Support issue, which specific features you are interested in.



Q: Are there any specific performance reasons for upgrading from 11.1 to the new version (Performance improvements - speed up execution that is)?
Q: Has performance been improved with v12 (assuming the same options are used)?

A: Yes, the version 12 compiler (Intel Fortran Composer XE) has numerous optimization improvements, even using the same options. Of course, your specific results may vary. As an example, loops with expressions with mixed data types (single and double precision, for example) are candidates for vectorization in version 12.



Q: How long is the Fortran standard going to be extended? Will there be an end to further standards? What is the long term strategy?

A: The Fortran standards committee is currently working on interpretations to the newly-approved Fortran 2008 standard. A revision, currently referred to as Fortran 2013, is being planned, but it is unclear what the scope of that will be. A majority of the committee wants it to be a minor update along the lines of Fortran 95.



Q: What are the plans for the editor enhancements? This has been asked a lot of times on the forum.
Q: Any plans for enhancements in Visual Studio (e.g., type viewer, module viewer), important for large projects
Q: Will there be regions so that we can hide program units that are not being developed?

A: We are working on enhancements to the Visual Studio environment, but can't comment on specifics.



Q: Does the new Intel Fortran Compiler support ubuntu 10.04?
A: Yes.



Q: Are there any plans for supporting the simple idb -gui debugging tool on OSX. Xcode is nice, but it requires a bit of investment in time to use.

A: We don't currently have plans for extending the idb debugger with a GUI interface at this time. There is a workaround to use the graphfical DDD front-end with Intel® Debugger. This workaround is documented in this forum post: http://software.intel.com/en-us/forums/showthread.php?t=60060.



Q: When will the IMSL libraries be included
A: Expect to see these available in late January or perhaps early February.



Q: When will defined type IO be added. I currently use namelists of temporary defined types.
A: We can't comment on a timeframe for this, but it will be there eventually.



Q: What about coarrays for Mac OS X*?
A: We have no plans for that at this time.



Q: Are there plans to revive Array Visualizer?
A: This is still being considered.



Q: How many multithreading methodologies does Intel Fortran Composer XE have? and What are they?
A: In the product itself, the compiler supports:
• Automatic parallelization (/Qparallel or –parallel), along with Guided Auto Parallelization to help tune the application for parallelism.
• OpenMP
• Coarrays
• DO CONCURRENT regions along with –parallel compiler option
Using external libraries, you can also use MPI or the operating system's threading calls.



Q: Any support for the old source browser option?
A: Work is being done in this area, but we can't comment further at this time.



Q: What is the difference between /fp:precise switch and the new improved switch?

A: As explained in the presentation, the /fp (-fp-model) switch is very different from the new option to select among alternate math libraries. The new options allow you to choose a variant of the math library that is faster but less precise, or one that delivers the same results on all processor types. The fp-model can affect how expressions are evaluated, and can disable/enable certain optimizations. The new libraries merely replace the default libimf library which contains intrinsics such as exp, sin, cos, etc. So these new libraries should be considered for use along with the –fp-model settings. For example, for very strict numeric results that you'd like to replicate on multiple cpu architectures: -fp-model exact –fimf-arch-consistency=true could be used together to insure the best possible results and consistency across many cpu generations. See the –fimf-* options. This could be a topic we revisit in a separate dedicated webinar next year.

For version 11.1, please see Intel Fortran Compiler 11.1 Release Notes

To get product updates, log in to the Intel® Software Development Products Registration Center.

For questions or technical support, visit Intel® Software Developer Support

Update 8 - December 2011
Intel® Fortran Composer XE 2011 for Linux*

• English

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English

Update 7 - October 2011
Intel® Fortran Composer XE 2011 for Linux*

• English
• Japanese

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English
• Japanese

Update 6 - September 2011
Intel® Fortran Composer XE 2011 for Linux*

• English

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English

Update 5 - July 2011
Intel® Fortran Composer XE 2011 for Linux*

• English

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English

Update 4 - May 2011
Intel® Fortran Composer XE 2011 for Linux*

• English

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English

Update 3 - March 2011
Intel® Fortran Composer XE 2011 for Linux*

• English
• Japanese

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English
• Japanese

Update 2 - January 2011
Intel ® Fortran Composer XE 2011 for Linux*

• English

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English

Update 1 - December 2010
Intel® Fortran Composer XE 2011 for Linux*

• English
• Japanese

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English
• Japanese

Initial Product Release - November 2010
Intel® Fortran Composer XE 2011 for Linux*

• English

Intel® Fortran Composer XE 2011 for Mac OS X*

• English

Intel® Visual Fortran Composer XE 2011 for Windows*

• English

> icl /Qpenmp testcase.cpp

> set KMP_AFFINITY=verbose,granularity=fine,proclist=[0,1,{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7}],explicit

> testcase.exe
......
OMP: Info #204: KMP_AFFINITY: decoding cpuid leaf 11 APIC ids.
OMP: Info #202: KMP_AFFINITY: Affinity capable, using global cpuid leaf 11 info
OMP: Info #154: KMP_AFFINITY: Initial OS proc set respected: {0,1,2,3,4,5,6,7}
OMP: Info #156: KMP_AFFINITY: 8 available OS procs
OMP: Info #157: KMP_AFFINITY: Uniform topology
OMP: Info #179: KMP_AFFINITY: 1 packages x 4 cores/pkg x 2 threads/core (4 total cores)
OMP: Info #206: KMP_AFFINITY: OS proc to physical thread map:
OMP: Info #171: KMP_AFFINITY: OS proc 0 maps to package 0 core 0 thread 0
OMP: Info #171: KMP_AFFINITY: OS proc 1 maps to package 0 core 0 thread 1
OMP: Info #171: KMP_AFFINITY: OS proc 2 maps to package 0 core 1 thread 0
OMP: Info #171: KMP_AFFINITY: OS proc 3 maps to package 0 core 1 thread 1
OMP: Info #171: KMP_AFFINITY: OS proc 4 maps to package 0 core 2 thread 0
OMP: Info #171: KMP_AFFINITY: OS proc 5 maps to package 0 core 2 thread 1
OMP: Info #171: KMP_AFFINITY: OS proc 6 maps to package 0 core 3 thread 0
OMP: Info #171: KMP_AFFINITY: OS proc 7 maps to package 0 core 3 thread 1
OMP: Info #147: KMP_AFFINITY: Internal thread 0 bound to OS proc set {0}
OMP: Info #147: KMP_AFFINITY: Internal thread 1 bound to OS proc set {1}
OMP: Info #147: KMP_AFFINITY: Internal thread 2 bound to OS proc set {2,3,4,5,6,7}
OMP: Info #147: KMP_AFFINITY: Internal thread 3 bound to OS proc set {2,3,4,5,6,7}
OMP: Info #147: KMP_AFFINITY: Internal thread 4 bound to OS proc set {2,3,4,5,6,7}
OMP: Info #147: KMP_AFFINITY: Internal thread 5 bound to OS proc set {2,3,4,5,6,7}
OMP: Info #147: KMP_AFFINITY: Internal thread 7 bound to OS proc set {2,3,4,5,6,7}
OMP: Info #147: KMP_AFFINITY: Internal thread 6 bound to OS proc set {2,3,4,5,6,7}
......

]$ export KMP_AFFINITY="verbose,granularity=fine,proclist=[0,1,{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7},{2,3,4,5,6,7}],explicit"