Intel® Cluster Toolkit Compiler Edition 4.0
Frequently Asked Questions (FAQ)
Copyright © 2007–2010 Intel Corporation
All Rights Reserved
Revision: 20100602
World Wide Web: http://www.intel.com
Disclaimer and Legal Information
INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL® PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER, AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
UNLESS OTHERWISE AGREED IN WRITING BY INTEL, THE INTEL PRODUCTS ARE NOT DESIGNED NOR INTENDED FOR ANY APPLICATION IN WHICH THE FAILURE OF THE INTEL PRODUCT COULD CREATE A SITUATION WHERE PERSONAL INJURY OR DEATH MAY OCCUR.
Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or characteristics of any features or instructions marked "reserved" or "undefined." Intel reserves these for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from future changes to them. The information here is subject to change without notice. Do not finalize a design with this information.
The products described in this document may contain design defects or errors known as errata which may cause the product to deviate from published specifications. Current characterized errata are available on request.
Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product order.
Copies of documents which have an order number and are referenced in this document, or other Intel literature, may be obtained by calling 1-800-548-4725, or by visiting Intel's Web Site.
Intel processor numbers are not a measure of performance. Processor numbers differentiate features within each processor family, not across different processor families. See http://www.intel.com/products/processor_number for details.
MPEG is an international standard for video compression/decompression promoted by ISO. Implementations of MPEG CODECs, or MPEG enabled platforms may require licenses from various entities, including Intel Corporation.
The software described in this document may contain software defects which may cause the product to deviate from published specifications. Current characterized software defects are available on request.
This document as well as the software described in it is furnished under license and may only be used or copied in accordance with the terms of the license. The information in this manual is furnished for informational use only, is subject to change without notice, and should not be construed as a commitment by Intel Corporation. Intel Corporation assumes no responsibility or liability for any errors or inaccuracies that may appear in this document or any software that may be provided in association with this document.
Except as permitted by such license, no part of this document may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without the express written consent of Intel Corporation.
Developers must not rely on the absence or characteristics of any features or instructions marked "reserved" or "undefined." Improper use of reserved or undefined features or instructions may cause unpredictable behavior or failure in developer’s software code when running on an Intel processor. Intel reserves these features or instructions for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from their unauthorized use.
BunnyPeople, Celeron, Celeron Inside, Centrino, Centrino Atom, Centrino Atom Inside, Centrino Inside, Centrino logo, Core Inside, FlashFile, i960, InstantIP, Intel, Intel logo, Intel386, Intel486, IntelDX2, IntelDX4, IntelSX2, Intel Atom, Intel Atom Inside, Intel Core, Intel Inside, Intel Inside logo, Intel. Leap ahead., Intel. Leap ahead. logo, Intel NetBurst, Intel NetMerge, Intel NetStructure, Intel SingleDriver, Intel SpeedStep, Intel StrataFlash, Intel Viiv, Intel vPro, Intel XScale, Itanium, Itanium Inside, MCS, MMX, Oplus, OverDrive, PDCharm, Pentium, Pentium Inside, skoool, Sound Mark, The Journey Inside, Viiv Inside, vPro Inside, VTune, Xeon, and Xeon Inside are trademarks of Intel Corporation in the U.S. and other countries.
* Other names and brands may be claimed as the property of others.
Copyright © 2007-2010, Intel Corporation. All rights reserved.
How do I use this FAQ? 02-Jun-2010
This FAQ for Intel® Cluster Toolkit Compiler Edition 4.0 is divided into the following hyperlinked topics:
Disclaimer and Legal Information
How do I use this FAQ? 02-Jun-2010
Product Capabilities 02-Jun-2010
Intel® MPI Library 15-Mar-2010
Intel® Debugger (IDB) 19-Feb-2010
Intel® Math Kernel Library 11-Jan-2010
Intel® Trace Analyzer and Collector 19-Feb-2010
Intel® MPI Benchmarks 07-Oct-2008
Uninstalling the Intel® Cluster Toolkit Compiler Edition 02-Jun-2010
Version Compatibility 02-Jun-2010
Each topic has a calendar date associated with it indicating as to when it was last modified. Within each topic, there are fact items and/or questions which are also dated according to when those fact items or questions were last revised.
Product Capabilities 02-Jun-2010
What is the Intel® Cluster Toolkit Compiler Edition? 02-Jun-2010
The Intel® Cluster Toolkit Compiler Edition 4.0 provides the high performance computing (HPC) with seven cluster-computing software tools bundled into a single integrated package with one license, one installation process, and one support service. The four component names in the toolkit are:
· Intel® C++ Compiler 11.1 Update 6
· Intel® Debugger 11.1 Update 6
· Intel® Fortran Compiler 11.1 Update 6
· Intel® MPI Benchmarks 3.2.1
· Intel® MPI Library 4.0
· Intel® Math Kernel Library 10.2 Update 5
· Intel® Trace Analyzer and Collector 8.0
The operating systems that the Intel® Cluster Toolkit Compiler Edition has been developed for are Linux* OS, Microsoft* Windows* Compute Cluster Server (also known as Microsoft Windows CCS*), and Microsoft* Windows* HPC Server 2008. HPC is an acronym for high-performance computing. In terms of the Intel Cluster Toolkit software for Windows*, consider references within this FAQ to Microsoft Windows CCS* OS and Microsoft* Windows* HPC Server 2008 OS as interchangeable.
What are the Intel® processor architectures that the Intel® Cluster Toolkit Compiler Edition will run on? 06-Oct-2006
The Intel® processor architectures are:
Intel® Pentium® 4 processor
Intel® Xeon® processor
Intel® Core™2 Duo processor (example of Intel® 64 (formerly Intel® EM64T) architecture)
Note: It is assumed that the processors listed above are configured into homogeneous clusters. Also, for the Microsoft Windows CCS* operating system, only processors based on the Intel® 64 architecture are supported.
What version of the Linux* operating system can I install the Intel® Cluster Toolkit Compiler Edition on? 05-Nov-2009
Presently Intel® Cluster Toolkit Compiler Edition is supported on:
OS Distributions |
IA-32 Architecture |
Intel® 64 Architecture |
|
32-Bit Applications |
64-Bit Applications |
||
Red Hat Enterprise Linux* 4.0 |
S |
S |
S |
Red Hat Enterprise Linux* 5.0 |
S |
S |
S |
SUSE Linux Enterprise Server* 10 |
S |
S |
S |
SUSE Linux Enterprise Server* 11 |
S |
S |
S |
What Microsoft Windows* operating system versions can I install the Intel® Cluster Toolkit Compiler Edition on? 17-Oct-2008
Presently Intel® Cluster Toolkit Compiler Edition is supported on:
Microsoft* Windows* Distributions |
IA-32 Architecture |
Intel® 64 Architecture |
IA-64 Architecture |
|
32-Bit Applications |
64-Bit Applications |
|||
Microsoft* Windows* Compute Cluster Server (Microsoft Windows CCS*) |
N/A |
S |
S |
N/A |
Microsoft* Windows* HPC Server 2008 |
N/A |
S |
S |
N/A |
What operating systems is Cluster OpenMP* for Intel® Compilers available on? 28-Apr-2008
Cluster OpenMP for Intel® Compilers is available on Intel® 64 and IA-64 architectures that run the Linux operating system.
How do I obtain a license for using Cluster OpenMP* for Intel® Compilers? 28-Apr-2008
As of Friday, May 2, 2008, a prerelease license for Cluster OpenMP will be available from the web-site:
Click on the web-page web-link to obtain a free license for the Cluster OpenMP library. Cluster OpenMP is an unsupported software product and may be used by customers through the prerelease End User License Agreement (EULA). Place this license in the directory:
/opt/intel/licenses
on your cluster system. It is also imperative that you and/or the system administrator set the environment variable INTEL_LICENSE_FILE to the directory path where the Intel software licenses will reside. For Bourne* Shell or Korn* Shell the syntax for setting the INTEL_LICENSE_FILE environment variable might be:
export INTEL_LICENSE_FILE=/opt/intel/licenses
For C Shell, the syntax might be:
setenv INTEL_LICENSE_FILE /opt/intel/licenses
What if I need to make inquiries about Cluster OpenMP*? 28-Apr-08
To make inquiries about Cluster OpenMP, visit the URL: http://whatif.intel.com. At the bottom of this landing page, there is a web link titled WhatIf Alpha Software Forums where you can review past questions, read what other people are working on, post a new inquiry, get support from product authors, and read the opinions of fellow users.
Intel® Cluster Tool Acronyms 15-Oct-2007
Acronym |
Definition |
ABI |
Application Binary Interface - describes the low-level interface an application program and the operating system, between an application and its libraries, or between component parts of an application. |
BLACS |
Basic Linear Algebra Communication Subprograms - provides a linear algebra oriented message passing interface for distributed memory computing platforms. |
BLAS |
Basic Linear Algebra Subroutines |
DAPL |
Direct Access Program Library - an Application Program Interface (API) for Remote Data Memory Access (RDMA). |
DFT |
Discrete Fourier Transform |
Ethernet |
Ethernet is the predominant local area networking technology. It is transports data over a variety of electrical or optical media. It transports any of several upper layer protocols via data packet transmissions. |
GB |
Gigabyte |
ICT |
Intel® Cluster Toolkit |
ICTCE |
Intel® Cluster Toolkit Compiler Edition |
IMB |
Intel® MPI Benchmarks |
IP |
Internet protocol |
ITA or ita |
Intel® Trace Analyzer |
ITAC or itac |
Intel® Trace Analyzer and Collector |
ITC or itc |
Intel® Trace Collector |
MPD |
Multi-purpose daemon protocol - a daemon that runs on each node of a cluster. These MPDs configure the nodes of the cluster into a “virtual machine” that is capable of running MPI programs. |
MPI |
Message Passing Interface - an industry standard, message-passing protocol that typically uses a two-sided send-receive model to transfer messages between processes. |
NFS |
The Network File System (acronym NFS) is a client/server application that lets a computer user view and optionally store and update file on a remote computer as though they were on the user's own computer. The user's system needs to have an NFS client and the other computer needs the NFS server. Both of them require that you also have TCP/IP installed since the NFS server and client use TCP/IP as the program that sends the files and updates back and forth. |
PVM* |
Parallel Virtual Machine |
RAM |
Random Access Memory |
RDMA |
Remote Direct Memory Access - this capability allows processes executing on one node of a cluster to be able to "directly" access (execute reads or writes against) the memory of processes within the same user job executing on a different node of the cluster. |
RDSSM |
TCP + shared memory + DAPL* (for SMP clusters connected via RDMA-capable fabrics) |
RPM* |
Red Hat Package Manager* - a system that eases installation, verification, upgrading, and uninstalling Linux packages. |
ScaLAPACK |
SCAlable LAPACK - an acronym for Scalable Linear Algebra Package or Scalable LAPACK. |
shm |
Shared memory only (no sockets) |
SMP |
Symmetric Multi-processor |
ssm |
TCP + shared memory (for SMP clusters connected via Ethernet) |
STF |
Structured Trace Format - a trace file format used by the Intel Trace Collector for efficiently recording data, and this trace format is used by the Intel Trace Analyzer for performance analysis. |
TCP |
Transmission Control Protocol - a session-oriented streaming transport protocol which provides sequencing, error detection and correction, flow control, congestion control and multiplexing. |
VML |
Vector Math Library |
VSL |
Vector Statistical Library |
Registration 21-Oct-2004
When using the Intel web-based registration process, I encounter diagnostic error messages such as “Invalid password format” or “Invalid challenge answer”. What am I doing wrong? 21-Oct-2004
For the password and challenge answer the minimum number of characters must be 8. For example, if the challenge question would be: “What is your mother’s maiden name?”, and your mother’s maiden name is 5 characters, you might want to increase the text length by doubling the maiden name length. E.g., suppose the maiden name is “Smith”, then enter “SmithSmith”.
Installation 29-Apr-2010
I tried installing the Intel® Cluster Toolkit Compiler Edition on the Ubuntu* operating system and I could not. What am I doing wrong? 30-Jun-2007
The Intel® Cluster Toolkit Compiler Edition should install just fine on the Ubuntu operating system, if you do not select the RPM install. You can do this by running the install.sh script and selecting the non-rpm choice, or you can use the --nonrpm option as part of the command-line for install.sh.
How come install packages for Intel® Cluster Toolkit Compiler Edition 3.1 or greater are so large with respect to previous releases? 18-Oct-2007
The installation packages for Intel® Cluster Toolkit Compiler Edition 3.1 or greater are much larger due to the integration of expanded Intel® Math Kernel Library functionality features. This will significantly increase the time required in doing a download of the Intel® Cluster Toolkit Compiler Edition installation package. Installing the cluster toolkit on every node of a cluster may also take substantially longer. Note that during the installation process the installer may need approximately 4 gigabytes of temporary disk storage to manage the intermediate installation files.
For the Linux* operating system, does the Intel® Cluster Toolkit Compiler Edition installer script which is called install.sh have any command-line options? 19-Feb-2010
The answer is yes. Simply type a shell command referencing install.sh such as:
install.sh --help | less
and you will see a list of options that look something like:
Usage: install.sh [options]
-h, --help print this message
-v, --version print version information
-s, --silent [FILE] run install silently, with settings in the configuration file
-d, --duplicate [FILE] run install interactively, record the user input into the configuration file
-t, --tmp-dir [FILE] set custom temporary folder
For the Linux* operating system, how much free space storage do I need in /tmp to complete an install of the Intel® Cluster Toolkit Compiler Edition? 19-Feb-2010
We recommend that you have at least two gigabytes of free space in /tmp when doing an install of the Intel® Cluster Toolkit Compiler Edition. Also, the installer script install.sh has the command-line options:
-t [FILE]
or
--tmp-dir [FILE]
which can direct the use of temporary intermediate storage to another disk partition. Again if an alternative disk partition is used during installation, approximately 4 gigabytes of temporary disk storage should be available to manage the intermediate installation files.
Are there any prerequisite tools that need to be present prior to installing the Intel® Cluster Toolkit Compiler Edition? 19-Feb-2010
For full functionality of the Intel® Cluster Toolkit Compiler Edition software package on Linux*, you will need to acquire additional software resources, if they are not already installed on your system:
· For all of the Intel® processor architectures running Red Hat Enterprise Linux 4.0, Red Hat Enterprise Linux 5.0, SUSE Linux Enterprise Server 10, SUSE Linux Enterprise Server 11, the version number on the Intel® Compilers should be 10.1 or greater.
· Perl* is Open Source software where the Perl programming language has been optimized for scanning arbitrary text files, extracting information from those text files, and generating printed reports based on that information. It is also a good language for many system administrative tasks. Perl is used in the Intel® Cluster Toolkit Compiler Edition by the Intel® Trace Collector. You can download it for free as source code, or as a pre-compiled binary distribution. The Perl source code is available for distribution by setting a Web browser at the URL: http://www.perl.com/pub/a/language/info/software.html. At this URL, click on the hyperlink titled, “Stable Production Release”.
· Python* is Open Source software where the high-level built in data structures in combination with dynamic typing and dynamic binding, make it very attractive for rapid application development, as well as for use as a scripting or glue language to connect existing software components together. Python is an interpreted, interactive, object-oriented programming language. The current release of Python can be retrieved from the URL: http://www.python.org/download. If Python needs to be installed on your system, then be sure to retrieve the current production version. The Python programming language is used by Intel® MPI Library.
· Expect* is a tool for automating interactive applications. You can download the expect software package from the following URL: http://expect.nist.gov/.
· PAPI* is an acronym for Performance API and it serves to gather information regarding performance counter hardware. Details can be found at the URL: http://icl.cs.utk.edu/papi/. The PAPI interface works in conjunction with Intel® Trace Analyzer and Collector.
In what instances do I need to use the expect software package? 30-Nov-2006
Within the “tar” package of the Intel Cluster Toolkit Compiler Edition, there is an expect shell script file called sshconnectivity.exp. This expect shell script will create or update a ~/.ssh directory on each node of the cluster beginning with the master node. If you do not have ssh-connectivity established on your Linux* cluster, the sshconnectivity.exp might be useful. As noted previously, you can download the expect software package from the following URL: http://expect.nist.gov/.
Where can I find TCL for Linux*? 21-Oct-2006
Any Linux implementation will probably have a TCL RPM where you should be able to download the TCL RPM off of that respective Linux vendor’s website. For example the following web site might be useful for downloading TCL RPMs: http://rpm.pbone.net/index.php3/stat/3/limit/2/srodzaj/1/dl/40/search/tcl/dist%5b%5d/47/dist%5b%5d/44/dist%5b%5d/41/dist%5b%5d/37/dist%5b%5d/31/dist%5b%5d/32/dist%5b%5d/1/dist%5b%5d/28/dist%5b%5d/2/dist%5b%5d/3/dist%5b%5d/4/dist%5b%5d/5/dist%5b%5d/7/dist%5b%5d/42/dist%5b%5d/36/dist%5b%5d/35/dist%5b%5d/11/dist%5b%5d/12/dist%5b%5d/49/dist%5b%5d/45/dist%5b%5d/34/dist%5b%5d/13/dist%5b%5d/14/dist%5b%5d/15/dist%5b%5d/17/dist%5b%5d/18/dist%5b%5d/46/dist%5b%5d/48/dist%5b%5d/33/dist%5b%5d/19/dist%5b%5d/20/dist%5b%5d/23/dist%5b%5d/24/dist%5b%5d/25/dist%5b%5d/26/dist%5b%5d/27/dist%5b%5d/29/dist%5b%5d/30/dist%5b%5d/38/dist%5b%5d/39/dist%5b%5d/40/dist%5b%5d/43/dist%5b%5d/50/field%5b%5d/1/field%5b.
I am trying to build an expect executable based on a download from http://expect.nist.gov/ and I am encountering the following error:
...
checking for Tcl private headers... checking for tclInt.h... no
configure: error: Can't find Tcl private headers
...
when running the configure command. What am I doing wrong? 30-Mar-2006
It is assumed that expect and Tcl are both stored in a path such as /usr/local/src. For example, the paths might be:
/usr/local/src/tcl8.4.12 (actual version may be different)
/usr/local/src/expect-5.43 (actual version may be different)
The header file tclInt.h is located in the generic folder under a Tcl installation path such as /usr/local/src/tcl8.4.12.The --with-tclinclude option should be used with the configure command to build expect. The configure command might look something like the following:
configure --with-tclinclude=/usr/local/src/tcl8.4.12/generic
For the Linux* operating system, I have established secure shell (SSH) connectivity on my cluster where each node has a public key. However, when I use any secure shell utility on the cluster, I am still prompted for a password. What am I doing wrong? 30-Nov-2006
Assuming that you have set up a NULL “passphrase” for your login account when using the ssh-keygen utility, for all systems in your cluster, and that you have distributed the public key for each node to all the other nodes (i.e., the public keys reside in the “~/.ssh/authorized_keys” file), then check the permissions of your home directory on your cluster. The “group” and “other” permission categories should only have read and execute accessibility for the home directory folder. If you activate a “write” permission setting for either the “group” or “other” categories, SSH will prompt you for the password. Also, the “~/.ssh” subdirectory should only have read, write, and execute permission for the login owner.
Note that the expect shell script sshconnectivity.exp will remove the write access capability on the group and other “permission categories” for the user’s home directory folder.
Regarding a Linux* installation, how do I install the Intel® Trace Analyzer on Microsoft Windows XP*? 19-Feb-2010
Suppose that the installation staging area for the Intel Cluster Toolkit Compiler Edition is:
/tmp/ict_staging_area
Within this directory after untarring the tar package you should see a folder that looks something like the following:
l_ict_<major>.<minor>.<update>.<package_num>
Recall that <major>.<minor>.<update>.<package_num> is a string such as:
b_4.0.0.035, where b is an acronym for beta
or
p_4.0.0.035, where p is an acronym for production
The <package_num> meta-symbol is a string such as 035. This string indicates the package number. Change to that directory with the shell command:
cd l_ict_<major>.<minor>.<update>.<package_num>
Within this folder, there is Microsoft Windows* installer package that should like something like w_itac_p_8.0.0.011.exe. Transfer this package along with the license file that was issued to you for the Intel® Cluster Toolkit Compiler Edition over to a Microsoft Windows XP* system and proceed to do an install of the Intel® Trace Analyzer on Windows. All of the features that are described for Intel® Trace Analyzer for Linux* are available for this Intel® Trace Analyzer for Microsoft Windows* application.
When I attempt to install the Intel® Cluster Toolkit Compiler Edition on my Linux* OS or Windows* OS system, and I have already pre-installed a software component (e.g. Intel® Trace Analyzer and Collector) of the Intel® Cluster Toolkit Compiler Edition with the exact same version number, the installer does not install that software component on any of the compute nodes. What am I doing wrong? 23-Apr-2010
On Linux* OS, if any software component of the Intel® Cluster Toolkit Compiler Edition is detected as having already been pre-installed on the head node, that software component will not be processed by the cluster toolkit installer. There is a similar problem on Windows* OS in the ‘Modify’ mode of the cluster toolkit installer. For Windows* OS, if some software component of the Intel® Cluster Toolkit Compiler Edition is pre-installed on the head node using the cluster toolkit installer, that software component will not be installed on the compute nodes of the cluster. For either Linux* OS or Windows* OS, if you already installed some of the software components on only the head node, and you want to install them on the other nodes using the cluster toolkit installer, you will need to uninstall such components from the head node manually before starting the cluster toolkit installer.
When I attempt to install the Intel® Cluster Toolkit Compiler Edition on my Linux* OS system using the command-line syntax:
./install.sh
the installer is generating the following message with respect to the Java* Runtime Environment (JRE):
Step no: 4 of 6 | Installation configuration > Missing Optional Pre-requisite
--------------------------------------------------------------------------------
There is one or more optional unresolved issues. It is highly recommended to fix
it all before you continue the installation. You can fix it without exiting from
the installation and re-check. Or you can quit from the installation, fix it and
run the installation again.
--------------------------------------------------------------------------------
Missing optional pre-requisite
-- No compatible Java* Runtime Environment (JRE) found
--------------------------------------------------------------------------------
1. Skip missing optional pre-requisites [default]
2. Show the detailed info about issue(s)
3. Re-check the pre-requisites
h. Help
b. Back to the previous menu
q. Quit
--------------------------------------------------------------------------------
Please type a selection or press "Enter" to accept default choice [1]:
What am I doing wrong? 29-Apr-2010
The Java* Runtime Environment is used by the Intel® Debugger graphical environment, which is a Java* application. On your Linux* OS platform, enter the value 2 after the prompt for the installation state shown above. This will generate the following diagnostic information:
Step no: 4 of 6 | Installation configuration > Missing Optional Pre-requisite
--------------------------------------------------------------------------------
This system does not appear to have a Java* JRE version 5.0 (also referred to as
1.5.0) installed. This may prevent operation of the Intel(R) Debugger GUI. For
further details, please refer to the System Requirement section of the product
Release Notes.
--------------------------------------------------------------------------------
1. Finish with prerequisites and continue installation [default]
2. Back to Pre-requisite summary dialog
h. Help
b. Back to the previous menu
q. Quit
--------------------------------------------------------------------------------
Please type a selection or press "Enter" to accept default choice [1]:
The instructions generated will indicate that a version of the Java* Runtime Environment is missing from your system. For the prompt above, enter the character “q” to quit the installation process. Make sure that the Java* Runtime Environment package is installed on your system. The directory path for where the Java* Runtime Environment may reside might be:
/usr/java
If you cannot find the Java* Runtime Environment library installation on your system, then visit the URL:
http://www.java.com/en/download/
to download the appropriate version of the Java* Runtime Environment. After doing the download, proceed to do an install of the Java* Runtime Environment on your system. You may need a system administrator to help you with this.
If you have located an existing and compatible Java* Runtime Environment library on your system, or you have proceeded to visit the URL above and did a download and installation, set your PATH environment variable to include the directory path to the Java* Runtime Environment library. The Bourne* and Korn* Shell syntax for setting the PATH environment variable might be something like the following:
export PATH=/usr/java/jre1.5.0_22/bin:$PATH
For C Shell, the syntax for setting the PATH environment variable might be something like:
setenv PATH /usr/java/jre1.5.0_22/bin:$PATH
After setting the PATH environment variable, repeat the installation process. The message regarding the missing Java* Runtime Environment library should go away.
Intel® MPI Library 15-Mar-2010
When I attempt to run an MPI application on Microsoft Windows CCS*, via the mpiexec command, I am seeing an error message that looks something like:
Unable to connect to 'clusternode1:8678',
sock error: generic socket failure, error stack:
MPIDU_Sock_post_connect(1200): unable to connect to clusternode1 on port 8678, exhausted all endpoints (errno -1)
MPIDU_Sock_post_connect(1247): unable to connect to clusternode1 on port 8678, No connection could be made because the target machine actively refused it. (errno 10061)
What am I doing wrong? 18-Oct-2007
Open a DOS Window on the head node of the Microsoft Windows CCS* cluster and type the command:
clusrun smpd -status
This may provide a report that might look something like the following:
-------------- clusternode1 returns 805982518 ---------------no smpd running on clusternode1
-------------- clusternode2 returns 671764790 --------------no smpd running on clusternode2
-------------- clusternode3 returns 604655926 --------------no smpd running on clusternode3
-------------- clusternode4 returns 873091382 --------------no smpd running on clusternod4
where the number of nodes providing a status will be a function of the size of the cluster. If you observe the above, then from the same DOS Window issue one of the following commands:
clusrun smpd -install
or
clusrun smpd -regserver
You should see messages that look something like the following:
------------------- clusternode1 returns 0 -----------------
Intel(R) MPI Library Process Manager, Intel installed.
------------------- clusternode2 returns 0 -----------------
Intel(R) MPI Library Process Manager, Intel installed.
------------------- clusternode3 returns 0 -----------------
Intel(R) MPI Library Process Manager, Intel installed.
------------------- clusternode4 returns 0 -----------------
Intel(R) MPI Library Process Manager, Intel installed.
where again the number of nodes providing an installation message will be a function of the size of the cluster. You should now be able to launch an mpiexec command as before.
On Microsoft Windows CCS*, when I first login and open up my first DOS panel using the “Build Environment for Intel® Cluster Toolkit Compiler Edition” shortcut, and launch an mpiexec command which looks something like the following:
mpiexec -n 2 -machinefile z:\cluster_file_share\machines.Windows my_mpi_prog.exe > my_mpi_prog.report 2>&1
nothing happens. What am I doing wrong? 22-Oct-2007
You may want to open a second DOS panel with the “Build Environment for Intel Cluster Toolkit Compiler Edition” shortcut, go to the working directory where the mpiexec command was issued, and enter the command:
type <my-output-file>
where <my-output-file> might be something like my_mpi_prog.report as shown in the mpiexec command above. If you see a prompt in the text file which looks something like:
User credentials needed to launch processes:
account (domain\user) [clusternode1\user001]:
Either press the return key or enter a valid user account. This should be followed by a prompt for a password which looks something like:
account (domain\user) [clusternode1\user001]: password:
Enter an appropriate password. From here the application should proceed with execution of the mpiexec launch command.
When I run "mpiifort test.f90" I get the following error:
ERROR: could not determine icc directory for driver: mpiifort
We are not using the Intel® C++ Compiler on our Linux* cluster system, nor does this error affect any of the other MPI compilation drivers. The Intel® Fortran Compiler works fine on its own. How can I fix this issue with Intel MPI Fortran compilation? 30-Nov-2006
The mpiifort compilation script calls a shell script named compchk.sh. A short term fix is to compile the application "test.f90" or "test.f" with "-nocompchk" as follows:
mpiifort -nocompchk test.f90
and
mpiifort -nocompchk test.f
Another solution is to install an Intel® C++ Compiler on the cluster system so as to resolve the apparent errors that are being detected by the shell script compchk.sh.
For the Linux* operating system, when compiling certain C++ applications using the Intel® MPI Library compilation driver mpiicpc, I encounter an error message that looks something like the following:
...
/opt/intel/ictce/3.2.2.035/impi/include/mpicxx.h(45): catastrophic error: #error directive: "SEEK_SET is #defined but must not be for the C++ binding of MPI"
#error "SEEK_SET is #defined but must not be for the C++ binding of MPI"
^
compilation aborted for my_mpi_program.cpp (code 4)
make: *** [my_mpi_program.o] Error 4
What am I doing wrong? 15-Jul-2009
For the application my_mpi_program.cpp try organizing the include files such that mpi.h comes before the stdio.h.
We have a cluster running the Linux* operating system, which has for example 3 nodes, where the mpd.hosts file contains:
$ cat mpd.hosts
node2
node3
The "master node", which is "node1" is not listed in the mpd.hosts file. The shell commands:
$ mpdboot -r ssh -n 2 -f ~/mpd.hosts
$ mpiexec -n 4 /bin/hostname
which are issued from "node1" reveal the following:
node1
node1
node2
node2
What am I doing wrong? 30-Nov-2006
By design, the "master node" (i.e., the node where the mpiexec command is launched from) needs to have a daemon running. If you want to run your application on node2 and node3, you can use the -host options with the mpiexec command as follows:
$ mpdboot -r ssh -n 3 -f ~/mpd.hosts
$ mpiexec -n 2 -host node2 /bin/hostname : -n 2 -host node3 /bin/hostname
This will run the hostname command on nodes node2 and node3 only, even though there are three mpd daemons total.
On Microsoft Windows CCS*, I am running an MPI application with an mpiexec command such as the following:
$ mpiexec -n 4 ./my_prog.exe
and I immediately obtain a command-line prompt. No error messages are issued and I do not obtain any results. What am I doing wrong? 15-Oct-2007
During the link phase of building the application executable on Microsoft Windows CCS*, try adjusting the stack size setting for the executable by using the /stack option as is illustrated with the following compile and link command:
$ icl /D_CRT_SECURE_NO_DEPRECATE /Femy_prog /I"%I_MPI_ROOT%"\em64t\include my_prog.c /link /stack:8000000 /LIBPATH:"%I_MPI_ROOT%\em64t\lib" impi.lib /NODEFAULTLIB:LIBCMTD.lib
For the example above, the value 8000000 sets the stack allocation in virtual memory to 8 million bytes. The default stack size is 1 MB.
On Microsoft Windows CCS*, I am running an MPI application with an mpiexec command such as the following:
$ mpiexec -n 4 ./my_prog.exe
and I am encountering a runtime error that looks something like the following:
forrtl: severe (29): file not found, unit 11, file c:\Windows\system32\input.dat
I do not have an explicit reference to the folder path c:\Windows\system32. I am using a mapped drive which serves as a file share to all nodes of the cluster. What am I doing wrong? 26-Jul-2008
Reconfigure your mpiexec command by using the -mapall option:
$ mpiexec -n 4 -mapall ./my_prog.exe
For Intel® MPI Library on Linux, how do I get online documentation for the various MPI functions? 15-Mar-2010
$ man <MPI-Function-Name>
where <MPI-Function-Name> is a meta-symbol for an actual MPI Standard function name. An example of this might be:
$ man MPI_Recv
Intel® Debugger (IDB) 19-Feb-2010
For Intel® MPI Library applications, what operating system or systems does the Intel® Debugger support? 10-Oct-2008
Intel® Debugger supports the debugging of Intel® MPI Library applications on Linux* OS.
To use the Intel® Debugger on Linux*, do I need to set any environment variables or set up any special source commands in my .bashrc or .cshrc file? 19-Feb-2010
Set the environment variable IDB_HOME to the folder path where the Intel Debugger executable, idb, resides. Also, you will want to source either idbvars.sh or idbvars.csh depending on which command-line shell you are using. For augmenting your .bashrc file, the Bourne* Shell or the Korn* Shell environment variable syntax for IDB_HOME and the sourcing of idbvars.sh simply source ictvars.sh. The syntax might be something like the following:
. /opt/intel/ictce/4.0.0.017/ictvars.sh
Alternatively, for ~/.cshrc the syntax might be something like:
source /opt/intel/ictce/4.0.0.017/ictvars.csh
The scripts ictvars.sh and ictvars.csh will manage the setting of environment variables that are associated with the Intel® Cluster Tools.
For the Linux* operating system, when I launch the Intel® Debugger through the Intel® MPI Library mpiexec command as follows:
mpiexec -idb –genv MPIEXEC_DEBUG 1 -n 4 ./testcpp
the mpiexec command seems to hang or go into an infinite loop. What am I doing wrong? 30-Nov-2006
If you do not have a ~/.rhosts file in your home directory (with permission bits set to 600), please create one. For each node in the computing cluster there should be a line in the ~/.rhosts with the following information:
<hostname as echoed by the shell command hostname> <your username>
As an example, suppose there are 8 nodes in the computing cluster with the following host names:
clusternode1
clusternode2
clusternode3
clusternode4
clusternode5
clusternode6
clusternode7
clusternode8
Assuming that the names listed above make up your cluster, they could be added to your ~/.rhosts file with the following general syntax:
<hostname as echoed by the shell command hostname> <your username>
For the list of nodes above and assuming that your login name is user01, the contents of your ~/.rhosts file might be:
clusternode1 user01
clusternode2 user01
clusternode3 user01
clusternode4 user01
clusternode5 user01
clusternode6 user01
clusternode7 user01
clusternode8 user01
The permission bit settings of ~/.rhosts should be set to 600 using the chmod command. The shell command for doing this might be:
chmod 600 ~/.rhosts
For the Linux* operating system, the run command for the Intel® Debugger seems to cause my MPI application to hang. What am I doing wrong? 30-Nov-2006
The "run" command is disabled in MPI debugging. To continue the execution of the MPI application use “cont”.
For the Linux* operating system, is there a rerun command when using the Intel® Debugger for an MPI application? 30-Nov-2006
Unfortunately, the rerun command is not yet supported within Intel Debugger. To restart, you will have to quit Intel Debugger and then re-enter the mpiexec command.
For the Linux* operating system, I have compiled the MPI application with the -g compiler option and the command-line for running the executable looks something like the following:
mpiexec -idb -n 4 ./test.exe
The Intel® Debugger launches with respect to the Intel® MPI Library command mpiexec, but the executable does not stop in main, but rather runs to completion. What am I doing wrong? 14-May-2009
The general syntax for using the Intel® Debugger with Intel® MPI Library is as follows:
mpiexec -idb –genv MPIEXEC_DEBUG 1 -n <number of processes> [other Intel MPI options] <executable> [arguments to the executable]
The environment variable MPIEXEC_DEBUG needs to be referenced so that MPI processes will suspend their execution to wait for the debuggers to attach to them. For the command-line example above, the –genv command-line option sets the environment variable MPIEXEC_DEBUG for all MPI processes. In general for Intel® MPI Library, the global environment variable command line switch –genv for mpiexec has the syntax:
–genv <environment variable> <value>
where <environment variable> is a meta-symbol that is a stand-in for a relevant environment variable, and <value> is a stand-in for setting an appropriate value for the environment variable that precedes the value argument.
Intel® Math Kernel Library 11-Jan-2010
For the Linux* operating system, I am trying to build ScaLAPACK executables from the directory path /opt/intel/ictce/4.0.0.035/mkl/tests/scalapack, where I am not explicitly setting the makefile variable LIBdir on the command-line. I encounter the following type of diagnostic error message:
...
mpiifort -nocompchk -o ../xslu_libem64t_intelmpi30_intel_noopt_lp64 psludriver.o psluinfo.o psgetrrv.o psmatgen.o pmatgeninc.o pslaschk.o pslafchk.o -L/shared/test_scalapack/scalapack/../../lib/em64t /shared/scratch/test_scalapack/scalapack/../../lib/em64t/libmkl_scalapack_lp64.a /shared/test_scalapack/scalapack/../../lib/em64t/libmkl_blacs_intelmpi_lp64.a -L/shared/test_scalapack/scalapack/../../lib/em64t /shared/test_scalapack/scalapack/../../lib/em64t/libmkl_intel_lp64.a -Wl,--start-group /shared/test_scalapack/scalapack/../../lib/em64t/libmkl_sequential.a /shared/test_scalapack/scalapack/../../lib/em64t/libmkl_core.a -Wl,--end-group -lpthread
ifort: error #10236: File not found: '/shared/test_scalapack/scalapack/../../lib/em64t/libmkl_scalapack_lp64.a'
ifort: error #10236: File not found: '/shared/test_scalapack/scalapack/../../lib/em64t/libmkl_blacs_intelmpi_lp64.a'
ifort: error #10236: File not found: '/shared/test_scalapack/scalapack/../../lib/em64t/libmkl_intel_lp64.a'
ifort: error #10236: File not found: '/shared/test_scalapack/scalapack/../../lib/em64t/libmkl_sequential.a'
ifort: error #10236: File not found: '/shared/test_scalapack/scalapack/../../lib/em64t/libmkl_core.a'
gmake[2]: *** [../xslu_libem64t_intelmpi30_intel_noopt_lp64] Error 1
gmake[2]: Leaving directory `/shared/test_scalapack/scalapack/source/TESTING/LIN'
gmake[1]: *** [exe] Error 2
gmake[1]: Leaving directory `/shared/test_scalapack/scalapack/source/TESTING'
gmake: *** [libem64t] Error 2
What am I doing wrong? 11-Jan-2010
For the makefile command-line, explicitly set the LIBdir variable to the path where the Intel® MKL resides. An example command-line might be something like:
gmake libem64t F=intel80 mpi=intelmpi30 LIBdir= /opt/intel/ictce/4.0.0.035/mkl/lib/em64t
Note that the make command above is applicable to Intel® 64 processor-based systems.
For Microsoft Windows CCS*, I am encountering the following link error message:
LINK : fatal error LNK1181: cannot open input file 'bufferoverflowu.lib'
when creating executables for ScaLAPACK and/or Cluster DFT. What am I doing wrong? 12-May-2008
Please source the .bat file:
vcvarsx86_amd64.bat
in your DOS command-line window where you are doing the Intel® Math Kernel Library experiments. This .bat file should be located in a bin subfolder within the Microsoft* Visual Studio* folder path and the DOS command for sourcing this file might look something like the following:
"C:\Program Files (x86)\Microsoft Visual Studio 8\VC\bin\x86_amd64\vcvarsx86_amd64.bat"
where the line above is contiguous.
Intel® Trace Analyzer and Collector 19-Feb-2010
I am trying to use the Intel® Trace Collector that comes with the Intel® Cluster Toolkit Compiler Edition with an MPI library other than Intel® MPI Library. I am seeing a link error, where there are unresolved symbols. What am I doing wrong? 07-Oct-2008
At the present time the Intel® Trace Collector that comes with the Intel® Cluster Toolkit Compiler Edition supports Intel® MPI Library and MPICH where the default library setting is Intel® MPI Library. To set the Intel® Trace Collector environment variables to support tracing of MPI applications that have been built with MPICH, use the following source command:
. .../itac/bin/itacvars.sh mpich
for Bourne Shell and for C Shell use:
source .../itac/bin/itacvars.csh mpich
where ... is the full installation path to where the itac folder resides on your system.
To reference Intel® MPI Library, you can replace the word mpich with the value mpi2 or mpi3 when sourcing itacvars.sh or itacvars.csh for the respective shell command-line examples above.
For the Linux* operating system, when I instrument the application test.cpp with the Intel® MPI Library compilation driver mpiicpc and with the Intel® Trace Collector using the following command line:
mpiicpc test.cpp -g -L${VT_LIB_DIR} -lVT ${VT_ADD_LIBS} -o testcpp_inst
I do not produce a trace file. What am I doing wrong? 20-Feb-2007
When compiling the C++ application test.cpp with the Intel® C++ Compiler (mpiicpc) and with Intel® Trace Collector requires an additional library called -lmpiic. The rule for this is that if one uses the MPI-2 C++ application programming interfaces (APIs) with the Intel® C++ Compiler, then Intel® Trace Collector cannot intercept the MPI calls as they are written. They have to be mapped to C function calls first, with the help of an MPI implementation-specific wrapper library, which with respect to the command-line, has to be placed in front of the Intel® Trace Collector library. The name of that wrapper library for the mpiicpc compilation driver is -lmpiic. Thus, for IA-32, and Intel® 64 processor-base architectures the compilation protocol should be:
mpiicpc test.cpp -g -lmpiic -L${VT_LIB_DIR} -lVT ${VT_ADD_LIBS} -o testcpp_inst
For a Linux* cluster, I am trying to use my keyboard to generate chart displays for Intel® Trace Analyzer 6.0 or greater and nothing appears. For example, I pressed the keys Ctrl+Alt+E on the left side of the keyboard which is suppose to generate the Event Timeline chart. What am I doing wrong? 30-Nov-2006
The keys may have been remapped by the various software interfaces on your system. If the Ctrl, Alt, and E keys on the left side of the keyboard do not generate an Event Timeline chart, please try using the Ctrl, Alt, keys on the right side of the keyboard along with the E key.
For the Linux* operating system, when I run Intel® Trace Collector on my Linux* cluster I am seeing something like the following diagnostic:
...
[0] Intel Trace Collector INFO: Writing tracefile test.stf in /tmp/test
[0] Intel Trace Collector WARNING: minimum clock increment 0.010000s is very high, please fix system setup to obtain better traces (see ITC FAQ)
...
What am I doing wrong? 30-Mar-2006
One can work around this problem by enabling the TSC timer. This is done by specifying the "tsc" option at boot time. One can view this problem description from the Red Hat Knowledgebase* at the URL: http://kbase.redhat.com/faq/FAQ_79_3728.shtm
For the Linux* operating system, when compiling and linking an MPI application on a cluster based on an Intel® 64 (formerly Intel® EM64T) architecture a message similar to the following is appearing when attempting to link with the Intel® Trace Collector Library (i.e. -lVT):
/usr/bin/ld: skipping incompatible /opt/intel/ictce/4.0.0.035/itac/lib/libVT.a
What am I doing wrong? 19-Feb-2010
The problem may be due to explicitly sourcing either the Bourne shell script mpivars.sh or the C Shell script mpivars.csh through a path such as:
/opt/intel/ictce/4.0.0.035/impi/bin
When the MPI application is compiled with one of the Intel MPI Library compilation drivers, a 32-bit version of object files are created. However, during the link step, the linkage editor detects an incompatibility between the 32-bit object files and the 64-bit Intel® Trace Collector Library (i.e., /opt/intel/ictce/4.0.0.035/itac/lib/libVT.a) and hence the error message. To correctly build 64-bit object files on Intel® 64 (formerly Intel® EM64T) architectures using the Intel® MPI Library compilation drivers, one needs to explicitly source either mpivars.sh or mpivars.csh in a directory path such as:
/opt/intel/ictce/4.0.0.035/impi/bin64
Another solution is to source one of the Intel® Cluster Toolkit Compiler Edition shell script files ictvars.sh or ictvars.csh that exist in a directory path such as:
/opt/intel/ictce/4.0.0.035/
The Bourne Shell source command might be something like:
. /opt/intel/ictce/4.0.0.035/ictvars.sh
The C Shell source command might be something like:
source /opt/intel/ictce/4.0.0.035/ictvars.csh
Intel® MPI Benchmarks 07-Oct-2008
For the Linux* operating system, when I use the following gmake command for the Intel® MPI Benchmarks:
gmake -f make_ict
I get the following error message:
make_ict:5: *** mpiicc is not defined through the PATH environment variable setting. Please try sourcing an Intel(r) Cluster Tools script file such as "mpivars.[c]sh" or "ictvars.[c]sh". Stop.
What am I doing wrong? 30-Nov-2006
Use of the makefile make_ict requires that mpivars.[c]sh or ictvars.[c]sh must be sourced. If you invoke the command above without having sourced one of these shell script files, a diagnostic message which looks like the above will appear.
How do I compile the Intel® MPI Benchmarks on Microsoft Windows CCS*? 07-Oct-2008
Within the Intel® MPI Benchmarks 3.2 folder configuration, there is a WINDOWS folder that contains Microsoft* Visual Studio* 2005 and 2008 project folders which allow you to use a pre-existing ".vcproj" project file in conjunction with Microsoft* Visual Studio* to build and run the associated Intel® MPI Benchmark application.
Within Microsoft Windows Explorer and starting at the Windows folder, you can go to one of the subfolders IMB-EXT_VS_2005, IMB-EXT_VS_2008, IMB-IO_VS_2005, IMB-IO_VS_2008, IMB-MPI1_VS_2005, or IMB-MPI1_VS_2008 and click on the corresponding ".vcproj" file and open up Microsoft Visual Studio*.
Uninstalling the Intel® Cluster Toolkit Compiler Edition 02-Jun-2010
For the Linux* operating system, does the Intel® Cluster Toolkit Compiler Edition uninstaller script which is called uninstall.sh have any options? 02-Jun-2010
The answer is yes. Simply type a shell command referencing uninstall.sh such as:
uninstall.sh --help | less
and you will see a list of options that look something like:
NAME
uninstall.sh - Uninstall Intel(R) Cluster Toolkit Compiler Edition for Linux* OS Version 4.0.0.
SYNOPSIS
uninstall.sh [options]
OPTIONS
--help Print this help and exit.
--temp-path=PATH
--temppath=PATH
Use specified folder for temporary files. By default, /tmp/install.XXXXXX folder is used.
--log-file=FILE
Write log to the specified file.
--single-node
--singlenode
Uninstall the product only from this node.
--delete-update=UPDATE_NUMBER
Delete update with the specified number.
--list-update
List all updates.
COPYRIGHT
Copyright(C) 1999-2010, Intel Corporation. All Rights Reserved.
How do I uninstall the Intel® Cluster Toolkit Compiler Edition on Microsoft Windows CCS*? 12-Jun-2008
Version Compatibility 02-Jun-2010
For the Linux* operating system, is Intel® Cluster Toolkit Compiler Edition 3.1 compatible with Intel® Cluster Toolkit 1.0, Intel® Cluster Toolkit 2.X, or Intel® Cluster Toolkit 3.X? 16-Jul-2007
Intel® Cluster Toolkit Compiler Edition 3.1 is not compatible with Intel® Cluster Toolkit 1.0. However, Intel® Cluster Toolkit Compiler Edition 3.1 is compatible with Intel® Cluster Toolkit 2.X and Intel® Cluster Toolkit 3.X.
For the Linux* operating system, is Intel® Cluster Toolkit Compiler Edition 3.1.1 compatible with Intel® Cluster Toolkit 1.0, Intel® Cluster Toolkit 2.X, or Intel® Cluster Toolkit 3.X? 29-Feb-2008
Intel® Cluster Toolkit Compiler Edition 3.1.1 is not compatible with Intel® Cluster Toolkit 1.0. However, Intel® Cluster Toolkit Compiler Edition 3.1.1 is compatible with Intel® Cluster Toolkit 2.X and Intel® Cluster Toolkit 3.X.
For the Linux* operating system, is Intel® Cluster Toolkit Compiler Edition 3.2 compatible with Intel® Cluster Toolkit 1.0, Intel® Cluster Toolkit 2.X, or Intel® Cluster Toolkit 3.X? 12-Jun-2008
Intel® Cluster Toolkit Compiler Edition 3.2 is not compatible with Intel® Cluster Toolkit 1.0. However, Intel® Cluster Toolkit Compiler Edition 3.2 is compatible with Intel® Cluster Toolkit 2.X and Intel® Cluster Toolkit 3.X.
For the Linux* operating system, is Intel® Cluster Toolkit Compiler Edition 3.2.1 compatible with Intel® Cluster Toolkit 1.0, Intel® Cluster Toolkit 2.X, or Intel® Cluster Toolkit 3.X? 10-Dec-2008
Intel® Cluster Toolkit Compiler Edition 3.2.1 is not compatible with Intel® Cluster Toolkit 1.0. However, Intel® Cluster Toolkit Compiler Edition 3.2.1 is compatible with Intel® Cluster Toolkit 2.X and Intel® Cluster Toolkit 3.X.
For the Linux* operating system, is Intel® Cluster Toolkit Compiler Edition 3.2.2 compatible with Intel® Cluster Toolkit 1.0, Intel® Cluster Toolkit 2.X, or Intel® Cluster Toolkit 3.X? 14-Jul-2009
Intel® Cluster Toolkit Compiler Edition 3.2.2 is not compatible with Intel® Cluster Toolkit 1.0. However, Intel® Cluster Toolkit Compiler Edition 3.2.2 is compatible with Intel® Cluster Toolkit 2.X and Intel® Cluster Toolkit 3.X.
For the Linux* operating system, is Intel® Cluster Toolkit Compiler Edition 4.0 compatible with Intel® Cluster Toolkit 1.0, Intel® Cluster Toolkit 2.X, or Intel® Cluster Toolkit 3.X? 02-Jun-2010
Intel® Cluster Toolkit Compiler Edition 4.0 is not compatible with Intel® Cluster Toolkit 1.0. However, Intel® Cluster Toolkit Compiler Edition 4.0 is compatible with Intel® Cluster Toolkit 2.X and Intel® Cluster Toolkit 3.X in regards to IA-32 and Intel® 64 architectures.