• Red Hat* Enterprise Linux* 4

• Ubuntu* 9 (for Intel MKL 10.3 - Intel MKL 10.3 update 3), 10.04, 11.04

• Debian* GNU/Linux 4&5, 6, Asianux* Server 3 and Turbolinux* 11)

• Red Hat Fedora* core 12, 13, 15

• Mac OS* X 10.5 (Leopard)

• Intel® C++ and Fortran Compiler 11.0 for Windows/Linux

• PGI* Workstation Complete version 10.8

• Intel MPI 3.0 and 3.1

Note: Ubuntu* 12.04, Red Hat Fedora* core 17 and PGI* Workstation Complete version 11.6 will be supported in Intel® MKL 11.0

To get more information on Intel MKL 10.3 system requirements refer to Intel® MKL 10.3 System Requirements Knowledge Base article.

If you have any concerns then please leave your comments at the end of this article, or discuss this on the Intel® MKL Forum.

a)   Copy sample source code: dgemmx.f and common_func.f from <mkl install directory /examples/blas/source> to your work directory, for example, /Users/yhu5/blas.
b)   Launch the Xcode application, typically found in the Developer > Applications folder.
c)   In Xcode Menu, select File > New > New Project.... , The New Project Assistant window opens. Select Application in the left pane. Select a template, for example, Command Line Tool, and click Next. Name your project, for example, MKL_dgemm_f, and select a type C. Click Next. Specify a directory for your project , For example, /Users/yhu5/blas. Click Create.
 

d)     Xcode creates the named project directory, with an .xcodeproj extension and several new source files like main.c etc.  Remove main.c, etc. file  and add dgemmx.f and common_func.f into ‘Source' project folder.

Each Xcode project has its own Project Editor window that displays project source files, targets, and executables.

2. Setting MKL Options in Xcode IDE

In the Xcode project, select the target and click Build Setting. Under the Build Settings tab, click All. Scroll down until you see IFORT_MKL, Select "Use threaded Intel® Math Kernel Library (or "Use non-thread Intel @ Math Kernal Library.)

3. Build and Run the Project

Select Product > Build. Or Click the  ‘Run' button on project tools bar, the project will build successfully. 
But may run with error message dyld: Library not loaded: libmkl_intel_lp64.dylib

(other compiler errors, like files is not found or Open file failed, please change the path of the files and the code in dgemmx.f, line 49, OPEN(UNIT=1, FILE='/Users/xx/xx/dgemmx.d',STATUS='OLD' so Xcode can find them.)

4) Set Run Environment

Open the Scheme Editor, click the Scheme button in the Project Editor Toolbar and select Edit Scheme, Under the Arguments tab, add DYLD_LIBRARY_PATH as an environment variable.
Set the Value as the full path to the Intel compiler's /lib directory and MKL directory
DYLD_LIBRARY_PATH=/opt/intel/composer_xe_2011_sp1.8.269/compiler/lib:

 /opt/intel/composer_xe_2011_sp1.8.269/mkl/lib

Click Run button again, you will get the result in Output box.

Please refer to Intel Compiler User and Reference Guide =>Building Applications with Xcode* IDE"
or "Configuring the Apple Xcode* Developer Software to Link with Intel MKL" section in MKL user guide for more details. 
And refer to "Linking Your Application with Intel® MKL" section in the Getting Started document to understand which libraries required by your code.

Operating System:

Mac OS*

Intel® Math Kernel Library (Intel® MKL) is a highly optimized, extensively threaded and thread-safe library of mathematical functions for engineering, scientific, and financial applications. Microsoft Office Excel* helps developers make more informed decisions, by using spreadsheets to analyze and share data. This article will show: how to create a Dynamic Link Library (DLL) from Intel MKL, call Intel MKL functions using a Microsoft Visual Basic for Applications (VBA), and interface with the DLL.

Building the Examples:   mklinExcel.zip

There are some ways of interfacing with third party software libraries from Excel which are well documented in Excel online help, like

How to create a user-defined function in Excel 2007 or Excel 2010

Developing User-Defined Functions for Excel 2007 and Excel 2010...

Create Custom Functions in Excel 2007

How to: Access DLLs in Excel - MSDN - Explore Desktop, Web ....

Calling from EXCEL through VBA do require that the called routine support STDCALL calling convention. Intel® MKL, which contains over hundreds math functions, support both CDECL and STDCALL interfaces (see, for example, MSDN, for description of calling conventions). Thus, the best way to call Intel® MKL functions from Excel to build a custom DLL based on Intel® MKL library which support STDCALL interface.

In this article we will use Microsoft Office Excel 2007 and Microsoft Visual Studio 2005 as example.

Prerequisites:
Install Intel MKL :  Intel Math Kernel Library product web page.
Install Microsoft office Excel 2007 and Microsoft Visual studio 2005

Step1: Build custom DLL by builder tools provided by Intel MKL

a.       Unzip the contents of the attached zip file to your work directory, for example C:\temp\mklinExcel\builder

 Enter the directory and edit the user_example_list text file with the functions you hope to use.

For example:

VDSQR 

The function performs element by element squaring of the vector. For example, calculate the square of one vector a= (a1, a2, ...) and vdsqr(a)=( a1^2,a2^2,...).

Search more functions in Intel® MKL Reference Manual  

b.       Run the build script (makefile) using nmake

Open a Microsoft Visual Studio command prompt or add the Microsoft.NET Framework to the PATH environment variable in the command prompt

And enter the directory and type the command

>nmake libia32 threading=sequential interface=stdcall export=user_example_list name=mkl_custom MKLROOT="C:\Program Files (x86)\Intel\composer XE 2011 SP1\mkl"

This will create a dll named mkl_custom.dll if success to build as below

Notes:

The makefile provides further explanation of the parameters in comments.

The builder tool is actually available in MKL install directory.   C:\Program Files (x86)\Intel\Composer XE 2011 SP1\mkl\tools\builder

Copy the builder directory (including all files) to your work directory, for example, C:\temp\builder\ and edit the makefile.

Change:

!if "$(interface)"=="std" || "$(interface)"=="stdcall" || "$(interface)"=="STD" || "$(interface)"=="STDCALL"

IFACE_LIB_32=mkl_intel_s.lib

IFACE_LIB_32_DLL=mkl_intel_s_dll.lib

!endif

Step 2. Build an example in Excel which call VDSQR() function

a.     Please open the mklDllTest.xls (in C:\temp\mklinExcel) in Excel 2007. 

Or you can create your own workshop in Excel and add command button as the steps. Add a button and assign a macro to it in a worksheet.

b.       Click Alt+F11 start Visual Basic Editor and edit Module1 to correct path of the custom DLL

'in this example, it is
Declare Sub VDSQR Lib "C:\temp\mklinExcel\builder\mkl_custom.dll" _

(ByRef n As Integer, ByRef a As Double, ByRef r As Double)

c.       Type your own VBA code in the sub procedure called CommandButton1_Click, as shown in the following code:

Private Sub CommandButton1_Click()

Dim n As Integer

Dim a() As Double

Dim b() As Double

Dim I As Integer

Dim Sum As Double

 

n = 3

ReDim a(n)

ReDim b(n)

 

For I = 0 To n

a(I) = I

Next I

 

Call VDSQR(n + 1, a(0), b(0))

 

Sum = 0

For I = 0 To n

Sum = Sum + b(I)

Next I

 

TextBox1.Text = "Answer 0^2+ 1^2 + 2^2 + 3^2 is " & Str(Sum)

 

End Sub

This will create a VBA to call mkl VDSQR function from Excel 2007. Please refer the below screenshots for more details

 

d. Click Alt+F11 back to Excel sheet. Click the "Call mklDLL vdSqr(n,a,r)" button, you will see the run result. 0^2+1^2 + 2^2 + 3^2 = 14.

Trouble shooting

•1.       Bad DLL calling convention (Error 49), please refer to the Excel on line help

The problem mainly is that your program is calling a routine in a DLL, but isn't using the StdCall calling convention.

Appendix C - References

• Intel® Math Kernel Library website
• Intel® Math Kernel Library User Forum
  

Forthcoming Intel® MKL 11.0 (Expected date of release: Autumn 2012)

•  Intel® MKL 11.0 will have backward incompatibility with Intel® MKL 10.2 update 3

•  Open MP static library on Microsoft* Windows

•  GMP* Arithmetic functions

•  Intel® MKL Reference Manual will be removed from product package

•  Intel® Pentium III will no longer be supported

•  Red Hat* EL4 and PGI* Fortran 10.x support will be dropped

•  PGI* Fortran 77 support will be removed

• a powerful N-dimensional array object
• sophisticated (broadcasting) functions
• tools for integrating C/C++ and Fortran code
• useful linear algebra, Fourier transform, and random number capabilities.

Besides its obvious scientific uses, NumPy can also be used as an efficient multi-dimensional container of generic data.

For more information on NumPy, please visit http://NumPy.scipy.org/

Version Information

This application note was created to help NumPy users to make use of the latest versions of Intel MKL on Linux platforms.

The instructions given in this articles apply to Intel MKL 10.3 and above and Intel Compiler 11.0 and above.

 
Step 2 - Downloading NumPy Source Code

The NumPy source code can be downloaded from:

http://NumPy.scipy.org/

Prerequisites

Intel MKL can be obtained from the following options:

Download a FREE evaluation version of the Intel MKL product.
Download the FREE non-commercial* version of the Intel MKL product.

All of these can be obtained at: Intel® Math Kernel Library product web page.

Intel® MKL is also bundled with the following products

Intel® Parallel Studio XE 2011
Intel Parallel Composer XE 2011
Intel Cluster Studio 2011

Step 3 - Configuration

Use the following commands to extract the NumPy tar files from the downloaded NumPy-x.x.x.tar.gz.

1.	$gunzip numpy-x.x.x.tar.gz 
2.	$tar -xvf numpy-x.x.x.tar 

 
The above will create a directory named numpy-x.x.x

Make sure that C++ and FORTRAN compilers are installed and they are in PATH. Also set LD_LIBRARY_PATH to your compiler (C++ and FORTRAN), and MKL libraries.

Step 4 - Building NumPy

Change directory to numpy-x.x.x
Create a site.cfg from the existing one
 
Edit site.cfg as follows:

Add the following lines to site.cfg in your top level NumPy directory to use Intel® MKL, if you are building on Intel 64 platform:

[mkl]
library_dirs = /opt/intel/composer_xe_2011_sp1.6.233/mkl/lib/intel64
include_dirs = /opt/intel/composer_xe_2011.sp1.6.233/mkl/include
mkl_libs = mkl_rt
lapack_libs =

If you are building NumPy for 32 bit, please add as the following

[mkl]
library_dirs = /opt/intel/composer_xe_2011_sp1.6.233/mkl/lib/ia32
include_dirs = /opt/intel/composer_xe_2011_sp1.6.233/mkl/include
mkl_libs = mkl_rt
lapack_libs = 

Modify cc_exe in numpy/distutils/intelccompiler.py to be something like:

self.cc_exe = 'icc -O3 -g -fPIC -fp-model strict -fomit-frame-pointer -openmp -xhost'

Here we use, -O3, optimizations for speed and enables more aggressive loop transformations such as Fusion, Block-Unroll-and-Jam, and collapsing IF statements, -openmp for OpenMP threading and -xhost option tells the compiler to generate instructions for the highest instruction set available on the compilation host processor. If you are using the ILP64 interface, please add -DMKL_ILP64 compiler flag.

Run icc --help for more information on processor-specific options, and refer Intel Compiler documentation for more details on the various compiler flags.

Compile and install NumPy with the Intel compiler: (on 64-bit platforms replace "intel" with "intelem")

python setup.py config --compiler=intel build_clib --compiler=intel build_ext --compiler=intel install

If you build NumPY for Intel64 bit platforms:

$export LD_LIBRARY_PATH=/opt/intel/composer_xe_2011_sp1.6.233/mkl/lib/intel64:/opt/intel/composer_xe_2011_sp1.6.233/compiler/lib/intel64:$LD_LIBRARY_PATH

If you build NumPY for ia32 bit platforms:

$export LD_LIBRARY_PATH=/opt/intel/composer_xe_2011_sp1.6.233/mkl/lib/ia32:/opt/intel/composer_xe_2011_sp1.6.233/compiler/lib/ia32:$LD_LIBRARY_PATH

It is possible that LD_LIBRARY_PATH causes a problem, if you have installed MKL and Composer XE in other directories than the standard ones. The only solution I've found that always works is to build Python, NumPy and SciPy inside an environment where you've set the LD_RUN_PATH variable, e.g:

$export LD_RUN_PATH=~/opt/lib:~/intel/composer_xe_2011_sp1.6.233/compiler/lib:~/intel/composer_xe_2011_sp1.6.233/mkl/lib/ia32

Note: We recommend users to use arrays with 'C' ordering style which is row-major, which is default than Fortran Style which is column-major, and this is because NumPy uses CBLAS and also to get better performance.

Appendex A: Example:

Please see below an example Python script for matrix multiplication that you can use Numply installed with Intel MKL which has been provided for illustration purpose.

import numpy as np
import time

N = 6000
M = 10000

k_list = [64, 80, 96, 104, 112, 120, 128, 144, 160, 176, 192, 200, 208, 224, 240, 256, 384]

def get_gflops(M, N, K):
	return M*N*(2.0*K-1.0) / 1000**3

np.show_config()

for K in k_list:
	a = np.array(np.random.random((M, N)), dtype=np.double, order='C', copy=False)
	b = np.array(np.random.random((N, K)), dtype=np.double, order='C', copy=False)
	A = np.matrix(a, dtype=np.double, copy=False)
	B = np.matrix(b, dtype=np.double, copy=False)

	C = A*B

	start = time.time()

	C = A*B
	C = A*B
	C = A*B
	C = A*B
	C = A*B

	end = time.time()

	tm = (end-start) / 5.0

	print "{0:4}, {1:9.7}, {2:9.7}".format(K, tm, get_gflops(M, N, K) / tm)

Appendix B: Performance Comparison

Please click Examples.py to download the examples for LU, Cholesky and SVD.



 

a. Open Visual Studio 2005/2008/2010.

b. On the main menu, select File > New > Project to open the New Project window.

c.  Select Project Types > Intel(R) Fortran > Console Application, then select Templates > Empty Project. When done, in the Name field, type <project name> for example, trig_mkl, and click OK. The New Project window closes.

The next steps are performed inside the Solution Explorer window. To open it, select View > Solution Explorer from the main menu.

2. Add sources of Intel MKL example to the project :

a. Right-click the Source Files folder under <project name> and select Add > Existing Item... from the drop-down menu. The Add Existing Item - <project name> window opens.

b. Browse to the Intel MKL example directory, for example, <mkl directory>\examples\pdettf\source. Select the   example file and supporting files with extension ".f" or ".f90" (Fortran sources). For example, select the d_trig_tforms_bvp.f90 file. For the list of supporting files in each example directory, click Add.The Add Existing Item - <project name> window closes, and the selected files appear in the Source Files folder in Solution Explorer.

c. Right-click the Header Files folder under <project name> and select Add > Existing Item... from the drop-down menu. The Add Existing Item - <project name> window opens.

d. Browse to the <mkl directory>\include directory. Select the header files, For example, select the mkl_dfti.f90 and mkl_trig_transforms.f90 files. Click Add. The Add Existing Item - <project name> window closes, and the selected files to appear in the Header Files folder in Solution Explorer.

The next step is to configuring properties of the project:

3.  Select the <project name>.

4.  On the main menu, select Project > Properties to open the <project name> Property Pages window.

5.  Set the Intel MKL include dependencies:

a.  Select Configuration Properties > Fortran > General. In the right-hand part of the window, select Additional Include Directories > > <Edit...>. The Additional Include Directories window opens.

b.  Type the Intel MKL include directory in quotes: "<mkl directory>\include". Click OK to close the window.

6. Select Configuration Properties > Fortran > Preprocessor. In the right-hand part of the window, select Preprocess Source File > Yes (default is No). This step is recommended because some examples require preprocessing.

7.   Set library dependencies: 

a.  Select Configuration Properties > Linker > General. In the right-hand part of the window, select Additional Library Directories > > <Edit...>. The Additional Library Directories window opens.

b.  Type the directory with the Intel MKL libraries in quotes, that is, "<mkl directory>\lib\<architecture>", where <architecture> is one of { ia32, intel64 }, for example: "<mkl directory>\lib\Intel64". Click OK to close the window.

c.  Select Configuration Properties > Linker > Input. In the right-hand part of the window, select Additional Dependencies and type the libraries required, for example, if <architecture> =intel64, type mkl_intel_lp64.lib mkl_intel_thread.lib mkl_core.lib libiomp5md.lib.

8.  In the <project name> Property Pages window, click OK to close the window.

9.  Some examples do not pause before the end of execution. To see the results printed in the Console window, set a breakpoint at the very end of the program or add the 'pause' statement before the last 'end' statement.

10. To build the solution, select Build > Build Solution.

11. To run the example, select Debug > Start Debugging. The Console window opens.

12.  You can see the results of the example in the Console window. If you used 'pause' statement to pause execution of the program, press Enter to complete the run. If you used a breakpoint to pause execution of the program, select Debug > Continue. The Console window closes.

Below each error is described in details:

0: no error

-1: Input inconsistency

• Incorrect stage number for PARDISO was called.
• Incorrect PARDISO calling sequence, e.g. run stage > 1 without initialization results in error reporting.
• Incorrect number of matrices to be solved was set (PARDISO can be used for solving several matrices with the same sparsity structure at once, taking into account that their maximum number was defined previously. Setting the number of matrices to be solved outside the range of 1≤ … ≤maxfct results in error).
• PARDISO checks the parameters at each stage for consistency with the parameters at the previous stages. Every disagreement results in error reporting.

-2: not enough memory

This error value is returned in the case of any problem with memory allocation inside PARDISO.

PARDISO messages:

"*** Error in PARDISO memory allocation: [STRUCTURE NAME], size to allocate: %d bytes"

"total memory wanted here: %d kbyte"

"symbolic (max): %d symbolic (permanent): %d"

"real(including 1 factor): %d"

It describes the issue that arises on allocation of STRUCTURE_NAME inside PARDISO. A additional information about current memory usages is also printed.

-3: reordering problem

Returned for any problem on a reordering stage (phase 11)

PARDISO messages:

"*** error PARDISO: reordering, symbolic factorization"

-4: zero pivot, numerical factorization or iterative refinement problem

Let us start with a citation of the Intel MKL manual:

Using phase =33 results in an error message (error =4, should be -4 * ) if the stopping criteria for the Krylow-Subspace iteration cannot be reached.

If phase= 23, then the factors L, U are recomputed for the matrix A and the error flag error=0 in case of a successful factorization. If phase =33, then error = -4 signals the failure

If the solver detects a zero or negative pivot for these matrix types, the factorization is stopped, PARDISO returns immediately with an error (error = -4) and iparm(30) contains the number of the equation where the first zero or negative pivot is detected.

The error returned in the case of any problem at the factorization stage (phase 22) or at the iterative refinement stage of solution.

PARDISO messages:

"*** Error in PARDISO: cgs error iparam(20) %d" – prints iparm(20) – see CG / CGS diagnostics in the Intel MKL manual

"*** error PARDISO: iterative refinement"

" contraction rate is greater than 0.9, interrupt" – rate of contraction is too small

"*** error PARDISO: iterative refinement"

" exceeds max. iteration number %d" - prints abs(iparm(8))

"*** Error in PARDISO: internal error, insufficient memory factorization" – looks like a problem at the factorization stage except for pivoting issues (see errors below)

"*** Error in PARDISO: zero or negative pivot, A is not SPD-matrix" – original matrix (almost) not SPD one (probably due to computer arithmetic inaccuracies)

"*** Error in PARDISO: zero pivot" – the same as above but for other matrix types

-5: unclassified (internal) error

This error value is not used currently and is reserved for the future use.

-6: preordering failed (matrix types 11, 13 only)

This error value is returned in the case of any problem at the stage of preparation for reordering (in matching algorithm).

PARDISO messages:

"*** Error in PARDISO: preordering failed after %d neqns out of %d"

"structure singular or input/parameter problem (matrix type 11,13)"

Looks like the message provides no meaningful information.

-7: diagonal matrix problem

PARDISO prints no messages.

-8: 32-bit integer overflow problem

This error value is returned on 32-bit architecture for big matrices when indices become greater than the maximal integer value on this platform.

PARDISO messages:

"*** error PARDISO: reordering, symbolic factorization"

-9: not enough memory for OOC

Let us start with a citation of the Intel MKL manual:

Note that if iparm(60) is equal to 1 or 2, and the total peak memory needed for strong local arrays is more than MKL_PARDISO_OOC_MAX_CORE_SIZE, the program stops with error -9. In this case, increase of MKL_PARDISO_OOC_MAX_CORE_SIZE is recommended.

This error value is returned when amount of memory available for PARDISO (defined by MKL_PARDISO_OOC_MAX_CORE_SIZE, by default 2000 Mb) is not enough to solve the current matrix. The issue can be resolved by increasing the value for available memory.

-10: problems with opening OOC temporary files

This error value is returned when PARDISO can’t create / open temporary files for storing OOC arrays, e.g. in the case of wrong permissions or when files were removed or blocked or not released after the previous steps.

-11: read/write problems with the OOC data file

This error value is returned when some problems appear in the process of working with files, e.g. in the case of no space left on device or problems with read / write operations because of algorithm issues.

* - the documentation error. Will be fixed in the version 10.3 Update3.

** - available memory means the RAM system's memory which is available at the moment of starting the calculations.