Intel®
Math Kernel Library 7.2.1 for
Linux*
Release Notes
Contents
Overview
Directory Structure
New in Intel® MKL 7.2.1
System Requirements
Installation
Known Limitations
Technical Support and Feedback
Related Products and Services
Disclaimer and Legal Information
The Intel® Math Kernel Library (Intel® MKL) provides developers of scientific, engineering and financial software with a set of linear algebra routines, discrete Fourier transforms and vectorized math and random number generation functions, all optimized for the latest Intel® Pentium® 4 processors, Intel® Xeon™ processors with Streaming SIMD Extensions 3 (SSE3) and Intel® Extended Memory 64 Technology (Intel® EM64T), and Intel® Itanium® 2 processors. Intel MKL provides linear algebra functionality with LAPACK (solvers and eigensolvers) plus levels 1, 2, and 3 BLAS offering the vector, vector-matrix, and matrix-matrix operations needed for complex mathematical software. For solving sparse systems of equations, Intel MKL now provides a direct sparse solver, for which two interfaces are provided: the PARDISO interface and the DSS interface. Intel MKL offers multidimensional discrete Fourier transforms (1D, 2D, 3D) with mixed radix support (not limited to sizes of powers of 2). Intel MKL also includes a set of vectorized transcendental functions (called the Vector Math Library (VML)) offering both greater performance and excellent accuracy compared to the libm (scalar) functions for most of the processors. The Vector Statistical Library (VSL) offers high performance, hand tuned vectorized random number generators for a number of probability distributions. Intel MKL offers multi-threading support using OpenMP* in addition to being a fully thread-safe library.
Performance on Intel® processors has improved on a number of functions, as noted below. In addition, this version of MKL also performs superbly on non-Intel (X86) processors allowing software developers to use a single library regardless of what hardware the customer uses.
Version 7.2.1 of Intel MKL introduces:
- Performance improvements and bug fixes.
For detailed information on these features, please refer to the "New in Intel® MKL 7.2.1"
section below.
The original versions of the BLAS from which that part of Intel MKL
was derived can be obtained from http://www.netlib.org/blas/index.html.
The original versions of LAPACK from which that part of Intel MKL was
derived can be obtained from http://www.netlib.org/lapack/index.html.
The authors of LAPACK are E. Anderson, Z. Bai,
C. Bischof, S. Blackford, J. Demmel, J. Dongarra, J. Du Croz,
A. Greenbaum, S. Hammarling, A. McKenney, and D. Sorensen.
The information below indicates the high level structure for Intel
MKL.
|
mkl721
|
Main directory
|
| |
mklnotes.htm
|
Release notes (this file)
|
| |
mkllic.htm
|
Intel MKL license
|
| |
redist.txt |
List of redistributable files
|
|
mkl721/doc
|
Directory for documents
|
| |
index.htm
|
Index to the Intel MKL documentation
|
| |
mklman.pdf
|
Intel MKL manual
|
| |
mkluse.htm
|
User notes for Intel MKL
|
| |
vmlnotes.htm
|
General discussion of VML
|
| |
vslnotes.pdf
|
General discussion of VSL
|
|
mkl721/examples
|
Source and data for examples
|
|
mkl721/include
|
Contains include files for both library routines and test and example
programs
|
|
mkl721/tests
|
Source and data for tests
|
|
mkl721/lib/32
|
Contains static libraries and shared objects for IA-32 applications
|
|
mkl721/lib/em64t
|
Contains static libraries and shared objects for applications running on processors with Intel EM64T
|
|
mkl721/lib/64
|
Contains static libraries and shared objects for the Itanium® 2 processor
|
|
mkl721/tools/builder
|
Contains tools for creating custom dynamically linkable libraries
|
|
mkl721/tools/environment
|
Contains shell scripts to set environment variables in the user shell
|
|
mkl721/tools/support
|
Contains a utility for reporting package ID and license key
information to Intel® Premier Support
|
New in Intel® MKL 7.2.1
- Performance improvements since Intel® MKL 7.2
- Improvements for the Intel® Itanium® 2 processor
- BLAS
- DGEMM: 1-3% improvement for TN and TT cases, and increased consistency of multi-threaded performance
- *TRMV: 20-300% improvement
- ZGERC, ZGERU: 20-30% improvement
- VML
- vdPowx: improved by 19 times for arguments 0^b (b constant).
- Improvements for the Intel® Pentium® 4 processor and Intel® Xeon™ processor with SSE3.
- BLAS
- DGEMM: 3-10% improvement
- CGEMM: 35-50% improvement
- ZGEMM: 35-50% improvement
- DTRSM: 10-60% improvement
- DSYMM: 3-10% improvement
- SDOT, CDOT(U,C): 200% improvement for problems fitting in L1 cache, 30-100% in L2 cache, and 20-40% in memory
- SASUM: 500% faster
- (C,CS)SCAL, (Z,ZD)SCAL: 15-25% faster
- SCASUM: 6% faster
- CGEMV: more than 200% faster
- ZGEMV: 26% faster
- (C,Z)HEMV: 15-60% faster
- DFT
- 1D DFT with double precision complex data: 8-14% faster
- LAPACK Optimizations
- DGESVD: 200-1000% improvement
- *EVD, *ORGQR: Threaded
- Improved the accuracy of single precision routines for linear least squares problems
- Other improvements
- Intel MKL is in the process of unifying how NaN’s are handled within the BLAS. This work is not complete so NaN responses may be different in version 7.2.1 than in previous versions.
- The memory manager has been repaired so that level 3 BLAS will work correctly on large NUMA systems. There is a limit on the maximum number of threads that can be created in the current version of Intel MKL. This will be resolved in a future version of Intel MKL. Contact customer support if this is a problem for your application.
- Built direct sparse solver (PARDISO) with -fpic flag for EM64T
- Intel MKL now supports the "medium" addressing model for its static archives on 64-bit Intel® Xeon processors (i.e., compiled with "-mcmodel=medium" flag
Recommended hardware: a PC, workstation or server, with Intel® Xeon™ processor, Intel® Xeon™ processors with SSE3 and Intel EM64T, Pentium 4 processor, or Itanium® 2 processor.
Software requirements
A supported OS (Intel MKL has been tested with the following):
- Red Hat* Linux* version 9.0 (on IA-32 systems only)
- Red Hat* EL 3.0
- SuSE* Linux* 8.2 (on IA-32 systems only)
- SuSE* Linux* Enterprise Server 8
- Intel EM64T: Red Hat Enterprise Linux 3.0 AS
- Intel EM64T: Red Hat Enterprise Linux 4.0 AS
- Intel EM64T: SuSE* Linux* Enterprise Server 9
A supported C or Fortran compiler (Intel MKL has been tested with the following):
- Intel® C++ Compiler versions 8.0 and 8.1 for IA-32 and Itanium® processors
- Intel® Fortran Compiler versions 8.0 and 8.1 for IA-32 and Itanium® processors
- Intel® C++ Compiler version 8.1 Extended Memory 64 Technology Edition
- Intel® Fortran Compiler version 8.1 Extended Memory 64 Technology Edition
- GNU compiler collection
Note: Intel Cluster MKL has parts which have Fortran interfaces, and
are Fortran in their data structures, and parts which have C
interfaces and have C data structures. The user notes file
(mkluse.htm in the doc directory) contains advice on how to link to
Intel Cluster MKL with different compilers.
To install the Intel MKL package on Linux*, use the following
instructions. The installation software installs the full Intel MKL
file set for all supported processors. See the Intel MKL
website for updates, when available.
- Use the tar command to extract the Intel MKL package in a directory to which you have write access
(e.g., tar -zxvf package.tar.gz).
- Become the root user and execute the install script in the
directory where the tar file was extracted by typing
"./install".
- The use of RPM necessitates root access to your system. If you do not have root access, you can find the extracted RPM package in a temporary directory. You may also contact customer support for work around information.
- If you are installing an evaluation version of Intel MKL you may
have to take a few steps to get the FLEXlm key license set up on your
system. Instructions are provided by the installer.
- The Intel® Performance Libraries products already installed
will be listed, followed by a menu of products to install which
includes:
- Intel® Math Kernel Library Version 7.2.1
- Select a package to install. Before the installation begins, you will be asked to enter a temporary directory in which the contents of the self-extracting installation file will be placed before actual installation begins. After installation, the files will still be located in this temp directory and must be deleted manually. All packages needed to use the product will also be installed. The recommended (default) installation directory is /opt/intel. In the directory you choose, a directory named mkl721 will be created and all files will be installed there. Any previous version of Intel MKL may remain installed when installing Intel MKL 7.2.1, but you will be required to remove the beta version of this software if you have it installed. Be sure to update your build scripts to point to the desired version of Intel MKL if you choose to keep multiple versions installed.
- The Intel MKL installation program uses RPM as the installation
vehicle. Some versions of RPM do not allow redirection of
installation. If the install program detects that you have a version
of RPM that does not allow redirection, you will be required to
install to the default directory.
- After installation, the packages installed will be redisplayed,
followed by a redisplay of the install menu. Enter 'x' to exit the
install script.
Three files, mklvars32.sh, mklvarsem64t.sh, and mklvars64.sh, will be placed in the tools/environment directory. These files can be used to set the INCLUDE and LD_LIBRARY_PATH environment variables in the current user shell.
Intel MKL uses Macrovision's* FLEXlm* electronic licensing
technology. License management should be transparent, but if you have
any problems during installation, please make sure a current license
file (*.lic) is located in
the same directory as the install file.
If you still have problems, please submit an issue to Intel®
Premier Support. See the "Technical
Support and Feedback" section of this document for
details.
Limitations to the sparse solver in Intel MKL 7.2.1:
- The default number of threads (when OMP_NUM_THREADS is not set) is
equal to the number of processors in the system. This differs from
the default OpenMP mode in Intel MKL (by default the number of threads
is set to one).
- Only statically linkable sparse solver library files will be
available with this release.
- Enhanced precision accumulation is implemented in long doubles (10
bytes real precision).
- Statistics output is not implemented (msglvl=1 will not deliver
statistics).
There are a number of limitations in the current implementation of
the set of DFT functions:
- The function DftiCopyDescriptor is not implemented.
- The function DftiGetValue is implemented with the following
restriction: the DFTI_FORWARD_ORDERING and DFTI_BACKWARD_ORDERING
parameters are not yet supported.
- Complex data is stored using the Fortran data type; real and
imaginary parts are adjacent.
- Modes DFTI_INITIALIZATION_EFFORT, DFTI_WORKSPACE, and
DFTI_TRANSPOSE are implemented only for the default case.
DFTI_FORWARD_SIGN can have the default value only and is not
changeable by the DftiSetValue function.
- DFTI_PRECISION, DFTI_DIMENSION, and DFTI_LENGTHS are settable only
through the DftiCreateDescriptor function and are not changeable by
the DftiSetValue function.
- Mode DFTI_FORWARD_DOMAIN can not have the value
DFTI_CONJUGATE_EVEN.
- 3D real DFT is not currently implemented.
- Modes DFTI_REAL_STORAGE and DFTI_CONJUGATE_EVEN_STORAGE can have
the default value only and are not changeable by the DftiSetValue
function (i.e., DFTI_REAL_STORAGE = DFTI_REAL_REAL and
DFTI_CONJUGATE_EVEN_STORAGE = DFTI_COMPLEX_REAL).
- Mode DFTI_COMPLEX_STORAGE can have the default value only and is
not changeable by the DftiSetValue function. In other words,
DFTI_COMPLEX_STORAGE is always DFTI_COMPLEX_COMPLEX.
When using the DFTs in Intel MKL it may be necessary to explicitly
link 'libm'. Please include '-lm' on your link line after any
reference to Intel MKL library files.
On Intel® processors with Intel® EM64T enabled, user programs
compiled with the GNU Fortran compiler (version 3.2.3) will likely get
incorrect results from those BLAS level 1 functions in Intel® MKL that
return single precision values. The GNU Fortran compiler expects REAL*4 values
in the first 8 bytes of the return register (just as a double precision value
would be represented) while the Intel® Fortran compiler expects REAL*4 values
in the first 4 bytes of the return register. The behavior of Intel MKL is
compatible with that of the Intel Fortran compiler.
Hyper-Threading Technology (HT Technology) is especially effective when each thread is performing different types of operations and when there are under-utilized resources on the processor. Intel MKL fits neither of these criteria as the threaded portions of the library execute at high efficiencies (using most of the available resources) and perform identical operations on each thread. You may obtain higher performance when using Intel MKL without HT Technology enabled.
DFT, VML, and VSL functions can not be used with Fortran 77
compilers.
Memory Allocation: In order to achieve better performance,
memory allocated by Intel MKL is not released. This behavior is by
design and is a one time occurrence for Intel MKL routines that
require workspace memory buffers. Even so, the user should be aware
that some tools may report this as a memory leak. Should the user
wish, memory can be released by the user program through use of a
function (MKL_FreeBuffers()) made available in Intel MKL
or memory can be released after each call by setting an environment
variable (MKL_DISABLE_FAST_MM) (see technical user notes
in the doc directory for more details). Using one of
these methods to release memory will not necessarily stop programs
from reporting memory leaks, and in fact may increase the number of
such reports should you make multiple calls to the library thereby
requiring new allocations with each call. Memory not released by one
of the methods described will be released by the system when the
program ends. The maximum number of buffers allocated in each thread
is 32. To avoid this restriction disable memory management as
described above.
On Red Hat* Enterprise Linux 3.0, in order to ensure that the
correct support libraries are linked, the environment variable
LD_ASSUME_KERNEL must be set:
For example: 'export LD_ASSUME_KERNEL=2.4.1'
Self Help and User Forums
A rich repository of self-help product information such as
tutorials, getting started tips, known product issues, product errata,
compatibility information and answers to frequently asked questions
can be found at the Intel®
Software Development Products Technical Support. It's a great
place to find answers quickly or to gain insight in using our products
effectively.
The Intel
MKL User Forum is the place to ask questions of and share
information with other users of Intel® MKL.
Submitting Issues
Your feedback is very important to us. To receive technical
support and product updates for the tools provided in this product you
need to register at the Intel®
Registration Center and click on "Create New Account".
For information about the Intel® MKL including FAQ’s, tips and
tricks, and other support information, please visit: http://support.intel.com/support/performancetools/libraries/mkl
Note: If you are having trouble registering or unable to
access your Premier Support account, contact developer.support@intel.com.
Please do not email your technical issue to developer.support@intel.com
as it is not a secure medium.
To submit an issue via the Intel® Premier Support website,
please perform the following steps:
- Ensure that Java* and JavaScript* are enabled in your browser.
- Go to https://premier.intel.com/.
- Type in your Login and Password. Both are case-sensitive.
- Click the "Submit Issues" button.
- Read the Confidentiality Statement and click the "I
Accept" button.
- Click on the "Go" button next to the
"Product" drop-down list.
- Click on the "Submit Issue" link in the left
navigation bar.
- Choose "Development Environment (tools,SDV,EAP)" from
the "Product Type" drop-down list.
- If this is a software or license-related issue choose
"Intel(R) MKL for Linux*"
from the "Product Name" drop-down list.
- Enter your question and complete the fields in the windows that
follow to successfully submit the issue.
Please follow these guidelines when forming your problem report or
product suggestion:
- Describe your difficulty or suggestion.
For problem reports
please be as specific as possible (e.g., including compiler and
link command line options), so that we may reproduce the
problem. Please include a small test case if possible.
- Describe your system configuration information.
Be sure to
include specific information that may be applicable to your
setup: operating system, name and version number of installed
applications, and anything else that may be relevant to helping
us address your concern.
Information on Intel® software development products is available
at
http://www.intel.com/software/products. Some of the related
products include:
- The
Intel® Software College provides interactive tutorials,
documentation, and code samples that teach Intel® architecture
and software optimization techniques.
- The
VTune™ Performance Analyzer allows you to evaluate how your
application is utilizing the CPU and helps you determine if there
are modifications you can make to improve your application's
performance.
- The
Intel® C++ and Fortran Compilers are an important part of
making software run at top speeds and fully support the latest Intel
IA-32 and Itanium® processors.
- The
Intel® Performance Library Suite provides a set of routines
optimized for various Intel® processors. The Intel® Math
Kernel Library, which provides developers of scientific and
engineering software with a set of linear algebra, fast Fourier
transforms and vector math functions optimized for the latest Intel
Pentium and Intel Itanium® processors. The Intel® Integrated
Performance Primitives consists of cross platform tools to build
high performance software for several Intel architectures and
several operating systems.
Disclaimer and Legal Information
The information in this manual is subject to change without notice and 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. This document and the software
described in it are furnished under license and may only be used or copied in
accordance with the terms of the license. No license, express or implied, by
estoppel or otherwise, to any intellectual property rights is granted by this
document. The information in this document is provided in connection with Intel
products and should not be construed as a commitment by Intel Corporation.
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. Intel products are not intended for use in medical, life saving,
life sustaining, critical control or safety systems, or in nuclear facility
applications.
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 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.
Intel, the Intel logo, Intel SpeedStep, Intel NetBurst, Intel NetStructure,
MMX, i386, i486, Intel386, Intel486, Intel740, IntelDX2, IntelDX4,
IntelSX2,Celeron, Intel Centrino, Intel Xeon, Intel XScale, Itanium, Pentium,
Pentium II Xeon, Pentium III Xeon, Pentium M, and VTune 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.
Copyright © Intel Corporation 2000 - 2005.