Forums
https://software.intel.com/en-us/view/forum-page-default/36938
enIntel® Math Kernel Library (Intel® MKL) 11.3 Beta and Intel® Parallel Studio XE 2016 Beta
https://software.intel.com/en-us/forums/topic/549590
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>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. <strong>Intel(R) Math Kernel Library ( Intel(R) MKL ) version 11.3 Beta </strong>is now available, as part of the <strong>Intel® Parallel Studio XE 2016 Beta</strong> program. </p>
<p>Follow the link in the <a href="https://software.intel.com/en-us/articles/intel-mkl-113-release-notes">Intel M</a><a href="https://software.intel.com/en-us/articles/intel-mkl-113-beta-release-notes">KL v.11.3 Beta Release Notes</a> to see what exciting new features are introduced!</p>
<p>To download the release, register for the Intel® Parallel Studio XE 2016 Beta through the <a href="http://softwareproductsurvey.intel.com/f/150460/2f82/">pre-Beta survey site</a>. </p>
<p>In addition to Intel MKL, the Intel Parallel Studio XE suite of products brings together many exciting new technologies along with improvements to Intel’s existing software development tools:</p>
<ul>
<li><em>Expanded Standards and Features </em><em>– </em>Scaling Development Efforts Forward<br />
Additional language support for C11 and C++14, Fortran 2008 Submodules and IMPURE ELEMENTAL, and C Interoperability from Fortran 2015, and OpenMP* 4.1 TR 3. New support for SIMD operator use with SSE integer types, Intel® Cilk™ Plus combined Parallel and SIMD loops, OpenMP* 4.0 user-defined reductions (C++ only), enhanced uninitialized variable detection (Fortran only), feature improvements to Intel’s Language Extensions for Offload, annotated source listings, and a new directory structure. All available in the <strong>Intel® C/C++ and Fortran Compiler 16.0 Beta</strong>.</li>
<li><em>Vectorization</em> – Boost Performance by Utilizing Vector Instructions/Units<br />
Vectorization Advisor identifies new vectorization opportunities as well as improvements to existing vectorization and highlights them in your code. It makes actionable coding recommendations to boost performance and estimates the speedup. Available in the new <strong>Intel® Advisor XE 2016 Beta</strong>!</li>
<li><em>Big Data Analytics</em> – Easily Build IA Optimized Data Analytics Application<br /><strong>Intel® Data Analytics Acceleration Library (DAAL) 2016 Beta</strong> will help data scientists speed through big data challenges with optimized IA functions</li>
<li>The <strong>Intel® Math Kernel Library 11.3 Beta</strong> introduces Inspector - Executor API for Sparse BLAS: a new two-stage API for sparse Matrix Vector Multiplication format, as well as MPI wrappers to support custom MPI Implementations</li>
<li>The <strong>Intel® Integrated Performance Primitives 9.0 Beta</strong> adds new APIs to support external threading – a feature which allows users to choose different threading approaches for the applications</li>
<li><em>Scalable MPI Analysis –</em> Fast & Lightweight Analysis for 32K+ Ranks<br /><strong>Intel® Trace Analyzer and Collector 9.1 Beta</strong> adds a new MPI Performance Snapshot feature for easy-to-use, scalable MPI statistics collection and analysis of large MPI jobs to identify areas for improvement</li>
<li>Enhanced OpenMP* analysis and MPI+OpenMP multi-rank analysis<br /><strong>Intel® VTune™ Amplifier 2016</strong> Beta adds OpenMP parallelization inefficiency, imbalance and work sharing analysis to tune for more efficient use of parallel regions. It also now supports multi-rank analysis of MPI compute nodes with or without OpenMP use.</li>
</ul>
<p>If you are ready to get started, follow this link to complete the pre-beta survey, register, and download the beta software:</p>
<p><a href="http://softwareproductsurvey.intel.com/f/150460/2f82/"><strong>Intel® Parallel Studio XE 2016 Pre-Beta Survey</strong></a><br />
For more details and information about this beta program, check out the <a href="https://software.intel.com/en-us/articles/intel-parallel-studio-xe-2016-beta">Intel® Parallel Studio XE 2016 Beta page</a>, which includes additional information in the FAQ and What’s New documents.</p>
<p>As a Beta tester, you’ll be expected to provide feedback to our development teams via Beta surveys and submissions at <a href="http://premier.intel.com/">Intel® Premier Support</a>.</p>
</div></div></div>Fri, 10 Apr 2015 10:00:25 +0000Gennady Fedorov (Intel)549590 at https://software.intel.comAnnouncing new product: Intel® Data Analytics Acceleration Library 2016 Beta
https://software.intel.com/en-us/forums/topic/541880
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>We are pleased to announce the release of <strong>Intel® Data Analytics Acceleration Library 2016 Beta</strong>! Intel® Data Analytics Acceleration Library is a C++ and Java API library of optimized analytics building blocks for all data analysis stages, from data acquisition to data mining and machine learning. It is a library essential for engineering high performance data application solutions. <a href="https://software.intel.com/en-us/articles/announcing-intel-data-analytics-acceleration-library-2016-beta">Click here </a>to see more.</p>
</div></div></div>Sat, 21 Feb 2015 01:01:23 +0000Zhang Z (Intel)541880 at https://software.intel.comIntel® Math Kernel Library 11.2 Update 3 is now available
https://software.intel.com/en-us/forums/topic/540818
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p> </p>
<p>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. <strong>Intel MKL 11.2 Update 3</strong> packages are now ready for download. Intel MKL is available as part of the <a href="http://software.intel.com/en-us/articles/intel-parallel-studio-xe/">Intel® Parallel Studio XE 2015 </a>. Please visit the <a href="http://software.intel.com/en-us/articles/intel-software-evaluation-center/">Intel® Software Evaluation Center</a> to evaluate this product.</p>
<p><a href="http://software.intel.com/en-us/articles/intel-mkl-112-bug-fixes">Intel® MKL 11.2 Update 3 Bug fixes</a></p>
<p><strong>What's New in Intel® MKL 11.2 Update 3 : </strong><a href="https://software.intel.com/en-us/articles/intel-mkl-112-release-notes">Release Notes</a></p>
</div></div></div>Sun, 08 Feb 2015 19:36:24 +0000Gennady Fedorov (Intel)540818 at https://software.intel.comIntel® MKL Cookbook Recipes
https://software.intel.com/en-us/forums/topic/516223
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>Intel MKL Users,</p>
<p>We would like to Introduce a new feature <a href="https://software.intel.com/en-us/mkl_cookbook"> Intel® MKL Cookbook</a>, an online Document with recipes for assembling Intel MKL routines for solving complex problems.Please give us your valuable feedback on these Cookbook recipes, and let us know if you want us to include more recipes and/or improve existing recipes.</p>
<p> Thank you for Evaluating </p>
<p>Intel MKL Team </p>
</div></div></div>Fri, 06 Jun 2014 16:41:05 +0000Sridevi (Intel)516223 at https://software.intel.comForum poll: Intel MKL and threading
https://software.intel.com/en-us/forums/topic/515557
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>Intel MKL users,</p>
<p>We would like to hear from you how you are using Intel MKL with threading. Do you use the parallel or sequential MKL? How do your multithreaded applications use MKL? We would appreciate you to complete a short survey. It takes no more than 5 minutes. Your feedback will help us to make Intel MKL a better product. Thanks!</p>
<p>Survey link: <a href="https://idz.qualtrics.com/SE/?SID=SV_5Bmh232m96WJK3b">https://idz.qualtrics.com/SE/?SID=SV_5Bmh232m96WJK3b</a></p>
<p> </p>
</div></div></div>Fri, 16 May 2014 18:33:27 +0000Zhang Z (Intel)515557 at https://software.intel.comLinking Intel MKL is easy
https://software.intel.com/en-us/forums/topic/283747
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p><pre class="brush: bash">icl prog.c /Qmkl</pre></p>
<p>or,</p>
<p><pre class="brush: bash">ifort prog.f /Qmkl</pre></p>
<p>That's the easiest way if you are using one of the latest Intel compilers on Windows*. There are similar <a href="http://software.intel.com/sites/products/documentation/studio/composer/en-us/2009/compiler_c/copts/common_options/option_mkl.htm">compiler options </a>for Linux* and Mac OS* X as well.</p>
<p>Another easy way is to use our new <a href="http://software.intel.com/en-us/articles/using-the-intel-mkl-dynamic-interface-for-windows/">dynamic linking model </a>which requires a link to just one library. Add mkl_rt.lib to your Windows* link line or add -lmkl_rt to your Linux* or Mac OS* X link line.</p>
<p>These new options willwork for the cases usedby most users. Those who use less common interfaces or threading models may still want to visit the <a href="http://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/">Link Line Advisor </a>to find the right set of libraries.</p>
</div></div></div>Thu, 19 May 2011 07:39:29 +0000TODD R. (Intel)283747 at https://software.intel.comMKL Cookbook Example 1 -- a Critique
https://software.intel.com/en-us/forums/topic/557295
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>The MKL Cookbook at <a href="https://software.intel.com/en-us/mkl_cookbook">https://software.intel.com/en-us/mkl_cookbook</a> has a set of interesting and useful examples of solving physics problems using MKL. After reading the first of the cookbook examples, which is nicely described at <a href="https://software.intel.com/en-us/node/507039">https://software.intel.com/en-us/node/507039</a>, and trying to verify the results given by the cookbook code at <a href="https://software.intel.com/en-us/mkl_cookbook_samples">https://software.intel.com/en-us/mkl_cookbook_samples</a>, I have some critical comments. I hope someone will view these comments as constructive criticism.</p>
<p>1. The brief description given at <a href="https://software.intel.com/en-us/mkl_cookbook">https://software.intel.com/en-us/mkl_cookbook</a> says "Finding an approximate solution to a nonlinear equation", which would suggest to most people that the example concerns solving a nonlinear equation in one variable. The example, in fact, is about solving a nonlinear PDE using finite-difference discretization, and solving the resulting system of nonlinear algebraic equations using Pardiso.</p>
<p>2. The nonlinearity is introduced through the temperature dependence of thermal conductivity, as described on <a href="https://software.intel.com/en-us/node/507039">https://software.intel.com/en-us/node/507039</a> under "Discussion" (two different but similar symbols are used for temperature -- v and the Greek letter 'nu' -- please fix this). To treat this nonlinearity, the discretized equations are solved iteratively, using the solution of each iteration to recompute the coefficients to be used in the next iteration. In principle, this is fine, and there are many applications where such iterations are necessary. However, in the specific problem chosen this iteration is unnecessary. Gustav Kirchoff showed in 1894 that using w = \int \mu dv as the dependent variable reduces the variable conductivity problem to an equivalent constant conductivity problem.</p>
<p>3. I added 'printf' statements to the code in file pardiso_nonlinear.c to print out the final solution for temperature. The solution is not only incorrect in the interior of the unit cube domain, but shows values of 1 at the faces of the cube instead of the specified value of 0. I have not debugged the 300-line code to find out what is wrong. Instead, I give below two different ways of solving the problem and the results, so that you (Intel/MKL group) can debug the C code in the cookbook.</p>
<p>4. The example code employs a mesh with 6<sup>3</sup> nodes, of which 4<sup>3</sup> are interior nodes with unknown values of the dependent variable, and solves for these unknowns by calling Pardiso with n = 216. Exploiting the symmetry of the problem enables us to recognize that there are only four distinct temperature values! Let us name these four values as follows:</p>
<p> x y z w</p>
<p> 0.2 0.2 0.2 p</p>
<p> 0.2 0.4 0.2 q</p>
<p> 0.4 0.4 0.2 r</p>
<p> 0.4 0.4 0.4 s</p>
<p>Then, the heat equation is represented by A w = b, where A =</p>
<p> 6 -3 0 0<br />
-1 5 -2 0<br />
0 -2 4 -1<br />
0 0 -3 3</p>
<p>and b = h<sup>2</sup>, with h = 0.2 being the mesh size, and w = [p q r s]<sup>T</sup>. The solution is w = [ 0.0196 , 0.0260, 0.0351, 0.0484]<sup>T</sup>. By reversing the Kirchoff transformation, we recover the solution for temperature as T = [ 0.0180, 0.0233, 0.0305, 0.0403]<sup>T</sup>.</p>
<p>5. Here is simple Fortran code to solve the problem using Gauss-Seidel iteration to solve the linear equations for w (denoted T in the code). To keep the code short, I have not exploited symmetry, using which would have reduced the number of variables. The output values agree with those given above in Item 4.</p>
<pre class="brush:fortran;">! Poisson equation in unit cube, with T = zero at faces.
program lapl3d
implicit none
integer, parameter :: nx=5, ny=5, nz=5 ! number of segments
real, dimension (0:nx,0:ny,0:nz) :: T = 0.0 ! Kirchoff flow temperature
real :: h=1.0/nx, dT, Tn
integer :: i,j,k,iter=0
do
dT=0
do i=1,nx-1
do j=1,ny-1
do k=1,nz-1
Tn = (T(i-1,j,k)+T(i+1,j,k)+ &
T(i,j-1,k)+T(i,j+1,k)+ &
T(i,j,k-1)+T(i,j,k+1) + h*h)/6.0 ! discrete Poisson eq.
dT = max(dT,abs(T(i,j,k)-Tn))
T(i,j,k) = Tn ! Gauss-Seidel
end do
end do
end do
iter=iter+1
if(mod(iter,10).eq.0)write(*,'(1x,I4,2x,E10.3)')iter,dT
if(iter.gt.500.or.dT.lt.1e-8)exit
end do
! recover temperatures from Kirchoff flow temperatures
do i=1,nx-1
do j=1,ny-1
write(*,10)i,j,(sqrt(0.2*T(i,j,k)+0.01)-0.1,k=1,nz-1)
end do
end do
10 format(1x,2I3,<nx-1>F10.5)
end program
</pre><p> </p>
<p> </p>
<p> </p>
</div></div></div>Tue, 05 May 2015 01:26:31 +0000mecej4557295 at https://software.intel.comGetting started with MKL+ScaLAPACK!
https://software.intel.com/en-us/forums/topic/557241
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>I am a student and this is my first time using MKL. I am interesting in applying a Cholesky factorization to a SPD huge matrix and then invert it. I have the chance to use many computers for that purpose, thus I am only interested in distributed solutions. For that reason, I would like to use MKL with ScaLAPACK.</p>
<p>Is there any introductory tutorial for my level? The <a href="https://software.intel.com/en-us/mkl_cookbook">Cookbook</a> seems to have some examples, but they do not seem introductory.</p>
<p>Does MKL provide a C interface for that? From <a href="https://software.intel.com/en-us/forums/topic/288028">this</a> old post, the answer seems to be no. But if so, why not just using only ScaLAPACK? I do not know Fortran and I thought MKL would get me out of trouble by providing a C interface for ScaLAPACK.</p>
</div></div></div>Mon, 04 May 2015 09:35:10 +0000Georgios S.557241 at https://software.intel.comCompile or executing problem with ifort on Mac OS X*
https://software.intel.com/en-us/forums/topic/557085
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>Hi,</p>
<p>I have just installed Intel® Parallel Studio XE Composer Edition for Fortran OS X* (m_fcompxe_2015.3.187.dmg), and I have just used it to compile my program on Mac OS 10.9.5 . It seems that there is no problem for compiling, creating the executable file (actually, there are still some warnings). However, error message came when I executed it... I really have no idea what is going on. Would you give me some comments on it ?</p>
<p>Here are some messages when I compile my program:</p>
<p>ifort -O2 -xSSSE3 -shared-intel -auto -openmp -c patom.f90 -L/opt/intel/mkl/lib/ -L/Users/mac/intel_ode/lib/intel64 -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -liomp5 -lpthread -liode_intel64 -lm -I/opt/intel/mkl/include/intel64 -I/Users/mac/intel_ode/include<br />
ifort -O2 -xSSSE3 -shared-intel -auto -openmp -c sub19.f90<br />
ifort -O2 -xSSSE3 -shared-intel -auto -openmp -o b.e patom.o sub19.o -L/opt/intel/mkl/lib/ -L/Users/mac/intel_ode/lib/intel64 -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -liomp5 -lpthread -liode_intel64 -lm -I/opt/intel/mkl/include/intel64 -I/Users/mac/intel_ode/include<br />
ipo: warning #11062: /Users/mac/intel_ode/lib/intel64/libiode_intel64.a is an archive, but has no symbols (this can happen if ar is used where xiar is needed)<br />
ld: warning: ignoring file /Users/mac/intel_ode/lib/intel64/libiode_intel64.a, file was built for archive which is not the architecture being linked (x86_64): /Users/mac/intel_ode/lib/intel64/libiode_intel64.a</p>
<p>Here are the error messages after I execute the executable file:</p>
<p>dyld: Library not loaded: libmkl_intel_lp64.dylib<br />
Referenced from: /Users/mac/programming/Mac/patom2.2/./b.e<br />
Reason: image not found<br />
Trace/BPT trap: 5</p>
<p> </p>
<p>p.s.: this program works when I compile it on the Linux using ifort.</p>
</div></div></div>Fri, 01 May 2015 02:12:29 +0000Belmiro C.557085 at https://software.intel.comPARDISO-Solving a large matrix containing many identical nonzero sparsity structure submatrices
https://software.intel.com/en-us/forums/topic/557010
<div class="field field-name-body field-type-text-with-summary field-label-hidden"><div class="field-items"><div class="field-item even" property="content:encoded"><p>Hi friends,</p>
<p>I am solving coupled problem analysis in FEM. I want to solve a matrix A.x = b, A could be</p>
<p>A = [A11, A12; A21, A22] and B has a size of [A11, A12]</p>
<p>A could contain 3x3, 4x4...nxn submatrices. The submatrices have the same nonzero sparsity structure (IA and JA). At the moment the solution may be reordering IA and JA from submatrices to a large matrix A to input into Pardiso Solver. However it will increase the storage memory for IA and JA.</p>
<p>It there any way to solve matrix A without reodering IA and JA? As I know multi right hand side method solves the only A matrix with different rhs, it can not apply for this case. Can anyone give me a hint?</p>
</div></div></div>Thu, 30 Apr 2015 08:26:09 +0000Long N.557010 at https://software.intel.com