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 that require maximum performance. Intel MKL 11.3 Update 1 packages are now ready for download. Intel MKL is available as part of the Intel® Parallel Studio XE and Intel® System Studio . Please visit the Intel® Math Kernel Library Product Page .

Intel® MKL 11.3 Bug fixes

New Features in MKL 11.3 Update 1

• Benchmarks:
  • Added new Intel® Optimized High Performance Conjugate Gradient (HPCG) Benchmark with second generation of Intel® Xeon Phi(TM) coprocessor support
  • Improved MP LINPACK benchmark organization by removing sub-optimal MP LINPACK binaries from the product
  • Improved MP LINPACK performance for Intel® Advanced Vector Extension 2 (Intel® AVX2) for 64-bit Intel MKL
• BLAS:
  • Improved BLAS Level 1 performance for Intel AVX2 and Intel® Advanced Vector Extensions 512 (Intel® AVX512)
  • Improved performance of BLAS Level 3 functions (S,D,C,Z)SYMM and (C,Z)HEMM with Intel® Threading Building Blocks (Intel® TBB) threading when left side specified and m>>n and when right side specified and n>>m
  • Improved parallel performance of ?GEMM for Intel AVX2 for 64-bit Intel MKL for matrices with moderate dimensions
  • Improved BLAS Level 3 performance for Intel AVX512
  • Improved performance of (S,D)GEMV for Intel AVX2 for 64-bit Intel MKL
  • Improved parallel performance of ?TRSM for Intel AVX2 for 64-bit Intel MKL
  • Improved ?NRM2 performance for Intel® Advanced Vector Extension (Intel® AVX) and Intel AVX2 for 32-bit and 64-bit Intel MKL
  • Fixed (S,D)GEMM issues for beta=0 cases affecting Intel AVX2 for 64-bit Intel MKL during multithreaded execution 
    For DGEMM, this affects matrices with N < 4000 and M/nthreads > 5004 
    For SGEMM, this affects matrices with N < 4000 and M/nthreads > 10008
  • Fixed (S,D)SYMM issues for beta=0 cases affecting Intel AVX2 for 64-bit Intel MKL during both multithreaded and sequential execution 
    For DSYMM, this affects matrices with M/nthreads > 5004 
    For SSYMM, this affects matrices with M/nthreads > 10008
• LAPACK
  • Introduced an Intel TBB threading layer providing Intel MKL composability for DSYEV and DPSTRF
  • Added C language LAPACK examples
  • Significantly improved performance of (D/S/C/Z)STEDC and (D/S/C/Z)YEVD functions for middle sized matrices with about 4K rows or columns
• Intel MKL PARDISO 
  • Added Intel TBB threading support
  • Added support for block compressed sparse row (BSR) matrix storage format
  • Added optimization for matrices with variable block structure
• FFTs
  • Significantly improved small 2D and 3D batched FFT performance on Intel Xeon Phi coprocessor and second generation Intel Xeon processor.
• Sparse BLAS:
  • Added Intel TBB threading support for matrix-vector multiplication for BSR and CSR matrices
  • Improved parallel performance of sparse matrix by sparse matrix multiplication for large matrices with the inspector-executor API
• Extended Eigensolver
  • Improved diagnostics in generalized eigenproblems for matrices which are not positive definite
• Random Number Generators
  • Improved performance of MCG59 and MCG31M1 basic generators and Gaussian ICDF distribution
• Summary Statistics
  • Improved performance of covariance, correlation and cross product for cases of larger p dimension
• Vector Mathematics
  • Improved performance for vsRint, vsNearbyInt, vsConj, and vdConj functions
  • Improved accuracy for the EP version of the vdDiv function, the HA version of the vsTan function, and the LA version of the vdTan function
  • Fixed incorrect setting of VML_STATUS_ACCURACYWARNING in the HA and LA complex functions vzAdd, vzSub, vzMul, vzMulByConj, and vzDiv
• ScaLAPACK
  • Added code examples demonstrating use of ScaLAPACK functionality
• Added mkl_version.h include file, which can be used to determine the version of Intel MKL at the time of compilation

Check out the Release Notes

Contents

• File:  w_mkl_11.3.1.146_online.exe

  Online Installer for Windows

• File: w_mkl_11.3.1.146.exe

  A File containing the complete product installation for Windows (32-bit/x86-64bit development)

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 that require maximum performance. Intel MKL 11.3 Update 1 packages are now ready for download. Intel MKL is available as part of the Intel® Parallel Studio XE and Intel® System Studio . Please visit the Intel® Math Kernel Library Product Page.

Intel® MKL 11.3 Bug fixes

New Features in MKL 11.3 Update 1

• Benchmarks:
  • Added new Intel® Optimized High Performance Conjugate Gradient (HPCG) Benchmark with second generation of Intel® Xeon Phi coprocessor support
  • Improved MP LINPACK benchmark organization by removing sub-optimal MP LINPACK binaries from the product
  • Improved MP LINPACK performance for Intel® Advanced Vector Extension 2 (Intel® AVX2) for 64-bit Intel MKL
• BLAS:
  • Improved BLAS Level 1 performance for Intel AVX2 and Intel® Advanced Vector Extensions 512 (Intel® AVX512)
  • Improved performance of BLAS Level 3 functions (S,D,C,Z)SYMM and (C,Z)HEMM with Intel® Threading Building Blocks (Intel® TBB) threading when left side specified and m>>n and when right side specified and n>>m
  • Improved parallel performance of ?GEMM for Intel AVX2 for 64-bit Intel MKL for matrices with moderate dimensions
  • Improved BLAS Level 3 performance for Intel AVX512
  • Improved performance of (S,D)GEMV for Intel AVX2 for 64-bit Intel MKL
  • Improved parallel performance of ?TRSM for Intel AVX2 for 64-bit Intel MKL
  • Improved ?NRM2 performance for Intel® Advanced Vector Extension (Intel® AVX) and Intel AVX2 for 32-bit and 64-bit Intel MKL
  • Fixed (S,D)GEMM issues for beta=0 cases affecting Intel AVX2 for 64-bit Intel MKL during multithreaded execution 
    For DGEMM, this affects matrices with N < 4000 and M/nthreads > 5004 
    For SGEMM, this affects matrices with N < 4000 and M/nthreads > 10008
  • Fixed (S,D)SYMM issues for beta=0 cases affecting Intel AVX2 for 64-bit Intel MKL during both multithreaded and sequential execution 
    For DSYMM, this affects matrices with M/nthreads > 5004 
    For SSYMM, this affects matrices with M/nthreads > 10008
• LAPACK
  • Introduced an Intel TBB threading layer providing Intel MKL composability for DSYEV and DPSTRF
  • Added C language LAPACK examples
  • Significantly improved performance of (D/S/C/Z)STEDC and (D/S/C/Z)YEVD functions for middle sized matrices with about 4K rows or columns
• Intel MKL PARDISO 
  • Added Intel TBB threading support
  • Added support for block compressed sparse row (BSR) matrix storage format
  • Added optimization for matrices with variable block structure
• FFTs
  • Significantly improved small 2D and 3D batched FFT performance on Intel Xeon Phi coprocessor and second generation Intel Xeon processor.
• Sparse BLAS:
  • Added Intel TBB threading support for matrix-vector multiplication for BSR and CSR matrices
  • Improved parallel performance of sparse matrix by sparse matrix multiplication for large matrices with the inspector-executor API
• Extended Eigensolver
  • Improved diagnostics in generalized eigenproblems for matrices which are not positive definite
• Random Number Generators
  • Improved performance of MCG59 and MCG31M1 basic generators and Gaussian ICDF distribution
• Summary Statistics
  • Improved performance of covariance, correlation and cross product for cases of larger p dimension
• Vector Mathematics
  • Improved performance for vsRint, vsNearbyInt, vsConj, and vdConj functions
  • Improved accuracy for the EP version of the vdDiv function, the HA version of the vsTan function, and the LA version of the vdTan function
  • Fixed incorrect setting of VML_STATUS_ACCURACYWARNING in the HA and LA complex functions vzAdd, vzSub, vzMul, vzMulByConj, and vzDiv
• ScaLAPACK
  • Added code examples demonstrating use of ScaLAPACK functionality
• Added mkl_version.h include file, which can be used to determine the version of Intel MKL at the time of compilation

Check out the Release Notes

Contents

• File:  l_mkl_11.3.0.101_online.exe

  Online Installer for Linux

• File: l_mkl_11.3.0.101.tgz

  A File containing the complete product installation for Linux (32-bit/x86-64bit development)

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 that require maximum performance. Intel MKL 11.3 Update 1 packages are now ready for download. Intel MKL is available as part of the Intel® Parallel Studio XE and Intel® System Studio . Please visit the Intel® Math Kernel Library Product Page .

Intel® MKL 11.3 Bug fixes

New Features in MKL 11.3 Update 1

• Benchmarks:
  • Added new Intel® Optimized High Performance Conjugate Gradient (HPCG) Benchmark with second generation of Intel® Xeon Phi coprocessor support
  • Improved MP LINPACK benchmark organization by removing sub-optimal MP LINPACK binaries from the product
  • Improved MP LINPACK performance for Intel® Advanced Vector Extension 2 (Intel® AVX2) for 64-bit Intel MKL
• BLAS:
  • Improved BLAS Level 1 performance for Intel AVX2 and Intel® Advanced Vector Extensions 512 (Intel® AVX512)
  • Improved performance of BLAS Level 3 functions (S,D,C,Z)SYMM and (C,Z)HEMM with Intel® Threading Building Blocks (Intel® TBB) threading when left side specified and m>>n and when right side specified and n>>m
  • Improved parallel performance of ?GEMM for Intel AVX2 for 64-bit Intel MKL for matrices with moderate dimensions
  • Improved BLAS Level 3 performance for Intel AVX512
  • Improved performance of (S,D)GEMV for Intel AVX2 for 64-bit Intel MKL
  • Improved parallel performance of ?TRSM for Intel AVX2 for 64-bit Intel MKL
  • Improved ?NRM2 performance for Intel® Advanced Vector Extension (Intel® AVX) and Intel AVX2 for 32-bit and 64-bit Intel MKL
  • Fixed (S,D)GEMM issues for beta=0 cases affecting Intel AVX2 for 64-bit Intel MKL during multithreaded execution 
    For DGEMM, this affects matrices with N < 4000 and M/nthreads > 5004 
    For SGEMM, this affects matrices with N < 4000 and M/nthreads > 10008
  • Fixed (S,D)SYMM issues for beta=0 cases affecting Intel AVX2 for 64-bit Intel MKL during both multithreaded and sequential execution 
    For DSYMM, this affects matrices with M/nthreads > 5004 
    For SSYMM, this affects matrices with M/nthreads > 10008
• LAPACK
  • Introduced an Intel TBB threading layer providing Intel MKL composability for DSYEV and DPSTRF
  • Added C language LAPACK examples
  • Significantly improved performance of (D/S/C/Z)STEDC and (D/S/C/Z)YEVD functions for middle sized matrices with about 4K rows or columns
• Intel MKL PARDISO
  • Added Intel TBB threading support
  • Added support for block compressed sparse row (BSR) matrix storage format
  • Added optimization for matrices with variable block structure
• FFTs
  • Significantly improved small 2D and 3D batched FFT performance on Intel Xeon Phi coprocessor and second generation Intel Xeon processor.
• Sparse BLAS:
  • Added Intel TBB threading support for matrix-vector multiplication for BSR and CSR matrices
  • Improved parallel performance of sparse matrix by sparse matrix multiplication for large matrices with the inspector-executor API
• Extended Eigensolver
  • Improved diagnostics in generalized eigenproblems for matrices which are not positive definite
• Random Number Generators
  • Improved performance of MCG59 and MCG31M1 basic generators and Gaussian ICDF distribution
• Summary Statistics
  • Improved performance of covariance, correlation and cross product for cases of larger p dimension
• Vector Mathematics
  • Improved performance for vsRint, vsNearbyInt, vsConj, and vdConj functions
  • Improved accuracy for the EP version of the vdDiv function, the HA version of the vsTan function, and the LA version of the vdTan function
  • Fixed incorrect setting of VML_STATUS_ACCURACYWARNING in the HA and LA complex functions vzAdd, vzSub, vzMul, vzMulByConj, and vzDiv
• ScaLAPACK
  • Added code examples demonstrating use of ScaLAPACK functionality
• Added mkl_version.h include file, which can be used to determine the version of Intel MKL at the time of compilation

Check out the Release Notes

Contents

• File:  m_mkl_online_2016.1.043.dmg

  Online Installer for OS X

• File: m_mkl_2016.1.043.dmg

  A File containing the complete product installation for OS X (32-bit/x86-64bit development)

We are introducing a Technical Preview of Intel® Distribution of Python*, with packages such as NumPy* and SciPy* accelerated using Intel MKL. Python developers can now enjoy much improved performance of many mathematical and linear algebra functions, with up to ~100x speedup in some cases, comparing to the vanilla Python distributions. The technical preview is available for everybody at no cost. Click here to register and download. For any questions, please jump onto the user forum. 

Hello.

I'm Using MKL Pardiso Sparse Solver (PARDISO).

Some time ago, I Using Intel Parallel Studio 2013 Version (MKL 11.2).

Recently, I'm Update Intel Parallel Studio 2016 (MKL 11.3).

This Version Support Cluster Sparse Solver. so I using this Function.

But Solving result is difference. (about 0.02)

Originally does the other result come out?

Or is there the condition getting the little more exact result?

One of our clients is getting the warning in the subject, followed by the program failing, running our software.  Any ideas what is causing this or how to resolve it?

The user is using a 32-bit version of Windows.

The processor is

Our software is built with “Intel(R) Parallel Studio XE 2015 Update 1 Composer Edition for Fortran Windows” .

I am running the Professional Edition for Fortran and C++ of Intel Parallel Studio XE 2016. This bug occurs in versions 16.0.0 and 16.0.1 at least.

When I discovered this bug, I was trying to source bin/compilervars.sh in order to set my LD_LIBRARY_PATH properly. After running this, my LD_LIBRARY_PATH contained a couple of suspicious entries:

daalvars.csh daalvars.sh /soft/spack/opt/spack/linux-x86_64aries/intel-16.0.1/compilers_and_libraries_2016.1.150/linux/daal/bin/lib/intel64_lin
daalvars.csh daalvars.sh /soft/spack/opt/spack/linux-x86_64aries/intel-16.0.1/compilers_and_libraries_2016.1.150/linux/daal/bin/../compiler/lib/intel64_lin

The installation directory I used was:

/soft/spack/opt/spack/linux-x86_64/binaries/intel-16.0.1

As you can see, something stripped the "/bin" out of the middle of my path. I tracked this bug down to another script that is sourced by compilervars.sh. If you look inside daal/bin/daalvars.sh, you'll see the following line:

local daal=${daal_bin/\/bin/}

This line is designed to strip the "/bin" from the end of a path. However, my path contains the word "/binaries". This bash replacement actually strips out the first occurrence of the word "/bin". In order to fix this bug, I changed the line to the following:

local daal=${daal_bin%\/bin}

The second problem with daalvars.sh is actually on the line above. The script added a "daalvars.csh daalvars.sh " to my path. To be fair, this is actually my fault. In my .bashrc, I created a function:

function cd {
    command -p cd "$@"&& ls
}

This function automatically runs "ls" every time I change directories. The problem with the script is that it runs "cd" to determine the directory:

local daal_bin=$(cd "$(dirname "${BASH_SOURCE[0]}")"&& pwd)

The easiest fix for me was to change the line into:

local daal_bin=$(command -p cd "$(dirname "${BASH_SOURCE[0]}")"&& pwd)

I know, it's definitely my fault for creating a "cd" function that obscures the system "cd". If you don't want to patch this then I totally understand. It's mainly the "/bin" stripping that I consider to be a bug.

I attached a patch that I created for this bug. In order to use it, download the path, place it in the daal/bin directory, and run:

patch -p1 < daalvars_patch.txt daalvars.sh

I couldn't figure out where to submit a bug report for this problem, so if you would like me to submit an official bug report just let me know.

Hi all,

I meet a strange problem in DataFitting Module while call df?InterpolateEx1D function. If I present the call back function for extrapolation , the result was wrong. But if set them to NULL, result is right.

My problem is:

x[] = {1, 1.5, 2.5, 3.5, 4.5, 5}, and y[] = {100, 120, 150, 170, 200, 200}, Interpolate method is stepwise constant interpolant. And I need found y value at x=4.

The following is the sample code. I hope that was something wrong in my code. Can you help me to explain it? thanks.

#include <stdio.h>
#include "mkl_df.h"


// fixed value extrapolate function parameters
struct fixvalue_extrap_params
{
    float value;
};

// fixed value extrapolate function
int fixvalue_extrap( MKL_INT64* n, MKL_INT64 cell[], float site[], float r[],
                     void *params )
{
    fixvalue_extrap_params *p = (fixvalue_extrap_params*)params;
    r[0] = p->value;
    return 0;
}


int main(int argc, char* argv[])
{
    float x[] = { 1., 1.5, 2.5, 3.5, 4.5, 5 };
    float y[] = { 100., 120., 150., 170., 200., 200. };
    int nx = 6;
    int ny = 1;
    const int nsite = 1;
    float site[nsite] = { 4 };
    float r[nsite];
    int cell[nsite];

    DFTaskPtr task;                     // Data Fitting task descriptor

    MKL_INT ndorder = 1;                    // size of array describing derivative
    MKL_INT dorder[] = { 1 };           // only value to calculate

    int errcode = 0;

    /***** Create Data Fitting task *****/
    errcode = dfsNewTask1D(&task, nx, x, DF_NON_UNIFORM_PARTITION, ny, y, DF_NO_HINT);

    /***** Edit task parameters for look up interpolant *****/
    errcode = dfsEditPPSpline1D(task, DF_PP_STD, DF_CL_STEPWISE_CONST_INTERPOLANT, 0, 0, 0, 0, 0, 0);

    /***** Interpolate using lookup method without extrapolate *****/
    errcode = dfsInterpolateEx1D(task, (DF_CELL | DF_INTERP), DF_METHOD_PP,
                                 nsite, site, 0, ndorder,
                                 dorder, 0, r, 0, cell,
                                 0, 0, 0, 0,
                                 0, 0, 0, 0);
    // this answer is right
    printf("interpolate reuslt for [%f] = [%f] in section %d\n", site[0], r[0], cell[0]);

    dfsInterpCallBack le_cb, re_cb;       // interpolation call backs
    void * le_params, *re_params;    // interpolation call backs parameters
    fixvalue_extrap_params le_fixparams, re_fixparams;

    le_cb = fixvalue_extrap;
    re_cb = fixvalue_extrap;
    le_fixparams.value = y[0];
    re_fixparams.value = y[nx-1];
    le_params = &le_fixparams;
    re_params = &re_fixparams;

    /***** Interpolate using lookup method *****/
    errcode = dfsInterpolateEx1D(task, (DF_CELL | DF_INTERP), DF_METHOD_PP,
                                 nsite, site, 0, ndorder,
                                 dorder, 0, r, 0, cell,
                                 le_cb, le_params, re_cb, re_params,
                                 0, 0, 0, 0);

    // this answer is wrong
    printf("interpolate reuslt with extrapolate for [%f] = [%f] in section %d\n", site[0], r[0], cell[0]);

    /***** Delete Data Fitting task *****/
    errcode = dfDeleteTask(&task);

    return 0;
}

 My output is :

interpolate reuslt for [4.000000] = [170.000000] in section 4
interpolate reuslt with extrapolate for [4.000000] = [150.000000] in section 4

Intel® Data Analytics Acceleration Library (Intel® DAAL) is the library of Intel® architecture optimized building blocks covering all stages of data analytics: data acquisition from a data source, preprocessing, transformation, data mining, modeling, validation, and decision making. Intel DAAL 2016 Update 2 packages are now ready for download. Intel DAAL is available as part of the Intel® Parallel Studio XE and also as part of the free (no-cost)  Community Licensing program of Intel performance libraries. Please visit the Intel® Data Analytics Acceleration Library Product Page.

What's New in Intel DAAL 2016 Update 2

• Improved numerical stability and error handling for EM GMM algorithm.
• Performance improvements for multi-class classifiers, SVM, kernel functions, Apriori, and ALS algorithms.
• Introduced support for Sorting algorithm in batch processing mode.
• Introduced support for CSR data layout format in the initialization phase of the KMeans algorithm.
• Bug fixes and other improvements in the library and its documentation. 

I would like to use a customized distance function instead of consine distance. I would like also make it a distributed algorithm for a large data set. Please advise how to do it with Intel DAAL.

By the way, where is the source code for java files like cosdistance/Method.java etc.? It would be very helpful to follow those code to write customized algorithms, which would make the library much more useful.

MLLib is able to do large distributed matrix multiplication. How to use this library to support such features?

The single-precision to double-precision (SP2DP) interface maps single-precision (for both real and complex types) BLAS and LAPACK routines in Intel® Math Kernel Library (Intel® MKL) to double-precision routines in Intel MKL. It does not change the behavior of double-precision routines.

The SP2DP interface was provided to make it easier to use Intel MKL BLAS and LAPACK for Fortran applications that follow the BLAS and LAPACK routine naming convention of the Cray* Fortran library. It was only applicable on Intel® 64 architectures. In the 64-bit Cray Fortran library, single precision data types are always mapped to 64-bit types since there are no double precision data types. So, for example, an SGEMM call in the Cray Fortran library is equivalent to a DGEMM call in Intel MKL. The SP2DP interface helped applications that follow this convention to use Intel MKL, which does differentiate 32-bit and 64-bit real and complex data types on Intel 64 architectures.

SP2DP interface was deprecated in Intel MKL 11.3 Update 2 and is removed from future releases starting with the Intel MKL 2017 Beta.

If you used the Intel MKL SP2DP interface, then now you must modify your code by renaming routine calls from {S,C}<FUNCTION> to {D,Z}<FUNCTION> and replace Intel MKL SP2DP interface library with Intel MKL ILP64 interface library in your link line. For example,

SGEMM -> DGEMM

CGEMM -> ZGEMM

For mixed-precision and other non-standard functions you must follow the table below.

This document provides a general summary of new features and important notes about the Intel® Data Analytics Acceleration library (Intel® DAAL) software product. Please see the following links to the online resources and documents for the latest information regarding Intel DAAL:

·         Intel® DAAL Product Page

·         Intel® DAAL 2017 Release Notes

·         Intel® DAAL 2017 Installation Guide

·         Intel® DAAL 2017 System Requirements

·         Intel® DAAL 2017 Getting Started

Links to documentation, help, and code samples can be found on the main Intel DAAL product page. For technical support visit the Intel DAAL technical support forum and review the articles in the Intel DAAL knowledge base.

Please register your product using your preferred email address. This helps Intel recognize you as a valued customer in the support forum and insures that you will be notified of product updates. You can read Intel's Online Privacy Notice Summary if you have any questions regarding the use of your email address for software product registration.

What's New in Intel® DAAL 2017 Beta

• Python programming language API
• Building blocks for support of neural networks based computations:
  • Layers
    • Two-dimensional convolutional
    • One-, two-, and three-dimensional max pooling
    • One-, two-, and three-dimensional average pooling
    • Fully connected
    • Dropout
    • Logistic
    • Hyperbolic tangent
    • Rectifier Linear Unit (ReLu)
    • Parametric Rectifier Linear Unit (pReLu)
    • Smooth Rectifier Linear Unit (smooth ReLu)
    • Softmax
    • Absolute value (abs)
    • Batch normalization
    • Local response normalization
    • Concat
    • Split
  • Neural network and its model
  • Optimization solvers
    • Stochastic gradient descent
    • Mini-batch stochastic gradient descent
    • Stochastic limited memory Broyden–Fletcher–Goldfarb–Shanno (lBFGS)
  • Objective functions
    • Mean squared error (MSE)
  • Tensor
• z-score normalization
• Bug fixes and other improvements in the library and its documentation

This document provides a general summary of new features and important notes about the Intel® Data Analytics Acceleration library (Intel® DAAL) software product. Please see the following links to the online resources and documents for the latest information regarding Intel DAAL:

·         Intel® DAAL Product Page

·         Intel® DAAL 2017 Beta Release Notes

·         Intel® DAAL 2017 Beta Installation Guide

·         Intel® DAAL 2017 Beta System Requirements

·         Intel® DAAL 2017 Beta Documentation Download

Links to documentation, help, and code samples can be found on the main Intel DAAL product page. For technical support visit the Intel DAAL technical support forum and review the articles in the Intel DAAL knowledge base.

Please register your product using your preferred email address. This helps Intel recognize you as a valued customer in the support forum and insures that you will be notified of product updates. You can read Intel's Online Privacy Notice Summary if you have any questions regarding the use of your email address for software product registration.

What's New in Intel® DAAL 2017 Beta

• Python programming language API
• Building blocks for support of neural networks based computations:
  • Layers
    • Two-dimensional convolutional
    • One-, two-, and three-dimensional max pooling
    • One-, two-, and three-dimensional average pooling
    • Fully connected
    • Dropout
    • Logistic
    • Hyperbolic tangent
    • Rectifier Linear Unit (ReLu)
    • Parametric Rectifier Linear Unit (pReLu)
    • Smooth Rectifier Linear Unit (smooth ReLu)
    • Softmax
    • Absolute value (abs)
    • Batch normalization
    • Local response normalization
    • Concat
    • Split
  • Neural network and its model
  • Optimization solvers
    • Stochastic gradient descent
    • Mini-batch stochastic gradient descent
    • Stochastic limited memory Broyden–Fletcher–Goldfarb–Shanno (lBFGS)
  • Objective functions
    • Mean squared error (MSE)
  • Tensor
• z-score normalization
• Bug fixes and other improvements in the library and its documentation
• Intel DAAL samples are now moved to online articles and removed from the installer packages.

Please see the following links to the online resources and documents for the latest information regarding Intel DAAL:

·         Intel® DAAL Product Page

·         Intel® DAAL 2017 Beta Release Notes

·         Intel® DAAL 2017 Beta Installation Guide

·         Intel® DAAL 2017 Beta System Requirements

·         Intel® DAAL 2017 Beta Documentation Download

These instructions assume a standalone installation of Intel® Data Analytics Acceleration Library (Intel® DAAL). If your copy of Intel® DAAL was included as part of one of our "suite products" (e.g., Intel® Parallel Studio XE) your installation procedure may be different than that described below; in which case, please refer to the readme and installation guides for your "suite product" for specific installation details.

Before installing Intel® DAAL, check the Product Downloads section of Intel® Registration Center to see if a newer version of the library is available. The version listed in your electronic download license letter may not be the most current version available.

The installation of the product requires a valid license file or serial number. If you are evaluating the product, you can also choose the "Evaluate this product (no serial number required)" option during installation.
If you have a previous version of Intel® DAAL installed you do not need to uninstall it before installing a new version. If you choose to uninstall the older version, you may do so at any time.

Installing Intel® DAAL on Windows* OS

You can install multiple versions of Intel® DAAL and any combination of 32-bit and 64-bit variations of the library on your development system.

These instructions assume you to have an Internet connection. The installation program will automatically download a license key to your system. If you do not have an internet connection, see the manual installation instructions below.

Interactive installation on Windows* OS

1. If you received the Intel® DAAL product as a download, double-click on the downloaded file to begin.
2. You will be asked to choose a target directory ("c:\Users\<Username>\Downloads\"  by default) in which the contents of the self-extracting setup file will be placed before the actual library installation begins. You can choose to remove or keep temporarily extracted files after installation is complete. You can safely remove the files in this "downloads"  directory if you need to free up disk space; however, deleting these files will impact your ability to change your installation options at a later time using the add/remove applet, you will always be able to uninstall.)
3. Click Next when the installation wizard appears.
4. If you agree with the End User License Agreement, click Next to accept the license agreement.
5. License Activation Options:
   • If you do have an Internet connection, skip this step and proceed to the next numbered step (below).
   • If you do not have an Internet connection, or require a floating or counted license installation, choose Alternative Activation and click Next; there will be two options to choose from:
     • Activate Offline:  requires a License File.
     • Use a License manager: Floating License activation
6. Enter your serial number to activate and install the product.
7. Activation completed. Click Next to continue.
8. If there is package from another update of Parallel Studio XE installed, you will be able to select update mode on Choose Product Update Mode dialog:
   1. I want to apply this update to the existing version.
      Using this option will result in the original version being replaced by the updated version.
   2. I want to install this update separate from the existing version.
      Using this option will result in the update being installed in a different location, leaving the existing version unchanged.
9. The Installation Summary dialog box opens to show the summary of your installation options (chosen components, destination folder, etc.). Click Install to start installation (proceed to step 15) or click Customize to change settings. If you select "Customize", follow steps 10-14.
   
   
10. In the Architecture Selection dialog box, select the architecture of the platform where your software will run.
11. In the Choose a Destination Folder dialog box, choose the installation directory. By default, it is C:\Program Files\IntelSWTools. You may choose a different directory. All files are installed into the Intel Parallel Studio XE 2017 subdirectory (if you chose I want to install this update separate from the existing version, all files are installed into the parallel_studio_xe_2017.0.xxx directory, where xxx is the package number).
12. Package contains components for integration into Microsoft Visual Studio*. You are able to select the Microsoft Visual Studio product(s) for integration on the Choose Integration target dialog box.
13. If Microsoft Compute Cluster Pack* is present, and the installation detects that the installing system is a member of a cluster, the dialog box will be shown which provides you an option to install the product on all visible  nodes of the cluster or on the current node only(by default installation on all visible nodes is performed).
14. The Installation Summary dialog box opens to show the summary of your installation options (chosen components, destination folder, etc.). Click Install to start installation.
15. Click Finish in the final screen to exit the Intel Software Setup Assistant.

Online Installation on Windows* OS

The default electronic installation package for Intel® DAAL for Windows now consists of a smaller installation package that dynamically downloads and then installs packages selected to be installed. This requires a working internet connection and potentially a proxy setting if you are behind an internet proxy. Full packages are provided alongside where you download this online install package if a working internet connection is not available.

Silent Installation on Windows* OS

Silent installation enables you to install Intel® DAAL on a single Windows* machine in a batch mode, without input prompts. Use this option if you need to install on multiple similarly configured machines, such as cluster nodes.

To invoke silent installation:

1. Go to the folder where the Intel® DAAL package was extracted during unpacking; by default, it is the C:\Program Files\Intel\Download\w_daal_2017.y.xxx folder.
2. Run setup.exe, located in this folder: setup.exe [command arguments]

If no command is specified, the installation proceeds in the Setup Wizard mode. If a command is specified, the installation proceeds in the non-interactive (silent) mode.

The table below lists possible values of  and the corresponding arguments.

Uninstalling Intel® DAAL for Windows* OS

Please see the following links to the online resources and documents for the latest information regarding Intel DAAL:

·         Intel® DAAL Product Page

·         Intel® DAAL 2017 Beta Release Notes

·         Intel® DAAL 2017 Beta Installation Guide

·         Intel® DAAL 2017 Beta System Requirements

·         Intel® DAAL 2017 Beta Documentation Download

System Requirements

The Intel® DAAL supports the IA-32 and Intel® 64 architectures. For a complete explanation of these architecture names please read the following article:
Intel Architecture Platform Terminology for Development Tools.

The lists below pertain only to the system requirements necessary to support application development with Intel® DAAL. Please review your compiler (gcc*, Microsoft Visual Studio* or Intel® Compiler Pro) hardware and software system requirements, in the documentation provided with that product to determine the minimum development system requirements necessary to support your compiler product.

Supported Operating Systems

• Windows 10* (IA-32/Intel® 64)
• Windows 8* (IA-32/Intel® 64)
• Windows 8.1* (IA-32/Intel® 64)
• Windows 7* (IA-32/Intel® 64) - Note: SP1 is required for use of Intel® AVX instructions
• Windows Server* 2008 R2 SP1 and SP2 (IA-32/Intel® 64)
• Windows HPC Server 2008 R2 (IA-32/Intel® 64)
• Windows Server* 2012 (IA-32/Intel® 64)
• Red Hat* Enterprise Linux* 6 (IA-32 / Intel® 64)
• Red Hat* Enterprise Linux* 7 (IA-32 / Intel® 64)
• Red Hat Fedora* core 20 (IA-32 / Intel® 64)
• Red Hat Fedora* core 21 (IA-32 / Intel® 64)
• SUSE Linux Enterprise Server* 11
• SUSE Linux Enterprise Server* 12
• Debian* GNU/Linux 6 (IA-32 / Intel® 64)
• Debian* GNU/Linux 7 (IA-32 / Intel® 64)
• Ubuntu* 12.04 (Intel® 64)
• Ubuntu* 13.04 (IA-32 / Intel® 64)
• Ubuntu* 14.04 LTS (IA-32/Intel® 64
• Ubuntu* 15.04 (IA-32 / Intel® 64)
• OS X* 10.11 (Xcode 7.0 and higher)

Note: Intel® DAAL is expected to work on many more Linux distributions as well. Let us know if you have trouble with the distribution you use.

Supported C/C++* compilers for Windows* OS:

• Intel® C++ Compiler 16.0 for Windows* OS
• Intel® C++ Compiler 17.0 for Windows* OS
• Microsoft Visual Studio* 2012 - help file and environment integration
• Microsoft Visual Studio* 2013 - help file and environment integration
• Microsoft Visual Studio* 2015 - help file and environment integration

Users of Intel® DAAL 2017 Beta can access the Developer Guide and Reference by downloading the attached ZIP archive. Unzip the archive to your local system, and click daal_ur_guides.htm to open it in your browser.

Intel® Parallel Studio XE 2017 Beta is introducing a lot of new features and a couple of new products. Please register for the Beta program, check out the new features introduced in Intel® Data Analytics Acceleration Library (Intel® DAAL) 2017 Beta and provide us feedback

Registration Link: Click the link here for Registration

What's new in Intel DAAL:

• Python programming language API
• Building blocks for support of neural networks based computations:
  • Layers
    • Two-dimensional convolutional
    • One-, two-, and three-dimensional max pooling
    • One-, two-, and three-dimensional average pooling
    • Fully connected
    • Dropout
    • Logistic
    • Hyperbolic tangent
    • Rectifier Linear Unit (ReLu)
    • Parametric Rectifier Linear Unit (pReLu)
    • Smooth Rectifier Linear Unit (smooth ReLu)
    • Softmax
    • Absolute value (abs)
    • Batch normalization
    • Local response normalization
    • Concat
    • Split
  • Neural network and its model
  • Optimization solvers
    • Stochastic gradient descent
    • Mini-batch stochastic gradient descent
    • Stochastic limited memory Broyden–Fletcher–Goldfarb–Shanno (lBFGS)
  • Objective functions
    • Mean squared error (MSE)
  • Tensor
• z-score normalization
• Bug fixes and other improvements in the library and its documentation
• Intel DAAL samples are now moved to online articles and removed from the installer packages.

Checkout Online Release notes for more information

Intel® Parallel Studio XE (Intel PSXE) Beta Articles and References

• Intel® Parallel Studio XE Beta Home Page
• Intel® Parallel Studio XE Release Notes
• Beta Registration link

Apache Spark* (http://spark.apache.org/) is a fast and general engine for large-scale data processing. Since its inception in 2014, Spark has become a widely adopted Big Data framework due to multiple advantages over Hadoop MapReduce. These advantages include: Fault-tolerant distributed data structures (Resilient Distributed Dataset), more operations available for data processing, ease-of-use (increased developer productivity), support for many types of clusters, and easy connection to many types of data sources. 

Spark comes with a stack of powerful libraries, including a popular machine learning library, MLlib (http://spark.apache.org/mllib/). MLlib is full of compute-intensive mathematical algorithms. However, the implementations in MLlib are not necessarily optimized for Intel Architectures. These days, Big Data infrastructures are predominantly built using Intel processors. It is therefore in many developers' interest to make Spark MLlib run faster on Intel based clusters. 

One way to make MLlib run faster is to replace MLlib algorithms with equivalent but more optimized implementations from the Intel® Data Analytics Acceleration Library (Intel® DAAL). This allows you to keep your workflow within Spark, so that at the same time your machine learning runs faster, you still enjoy Spark's other advantages, 

Intel DAAL is a software solution for developing data applications in C++, Java, or Python. The library provides a set of optimized building blocks that can be used in all stages of the data analytics workflow. These building blocks include data mining methods such as basic statistical moments, Principle Component Analysis, associating rule mining, anomaly detection, etc.; and supervised and unsupervised machine learning methods such as linear regression, classification, Support Vector Machine, clustering, etc.

See the attached presentation for a recipe on how to build faster data applications on Spark using Intel DAAL. A companion ZIP archive contains code samples discussed in the presentation. Download and unzip the archive, and build the samples with these steps:

1. Edit pom.xml to set the correct path for 'daal.jar' on the build system. Let DAALROOT be an environment variable pointing to your Intel DAAL installation location, then 'daal.jar' is in $DAALROOT/daal.jar.
2. Build the samples with Maven (version 3.3 and above is required): 
   
   mvn clean package -DskipTests

To learn more about Intel DAAL, please visit the product page: https://software.intel.com/en-us/intel-daal

If you have any questions, please ask them on our user forum: https://software.intel.com/en-us/forums/intel-data-analytics-acceleration-library 

I'm trying to install daal library in python and I'm getting the error:

c:\mingw\bin\gcc.exe -mdll -O -Wall -I./daal/include "-IC:\Program Files (x86)\IntelSWTools\compilers_and_libraries_2017.0.048\windows\daal/include" -IC:\Python27\lib\site-packages\numpy\core\include -IC:\Python27\include -IC:\Python27\PC -c daal\wrp\_algorithms__kernel_function__linear.cpp -o build\temp.win32-2.7\Release\daal\wrp\_algorithms__kernel_function__linear.o /w
gcc: /w: No such file or directory

I'm using python 2.7, cython and numpy are installed (I'm using win 64).