Recent posts
https://software.intel.com/en-us/recent/790722
encluster_sparse_solver discrepancy
https://software.intel.com/en-us/forums/topic/535078
<p>Hello,</p>
<p>I'm trying to solve a general system with CPARDISO. When using two processes, there is no issue if I don't use the coefficient array during the solution phase. When using only one process, then I get a segmentation fault. Could you give me some insight into this issue, please ? Thank you in advance.</p>
<p>$ mpicxx -cxx=icpc cl_solver_unsym_complex_c.cpp -lmkl_intel_thread -lmkl_core -lmkl_intel_lp64 -liomp5 -std=c++11<br />
$ mpirun -np 1 ./a.out<br />
$ echo $?<br />
11<br />
$ mpirun -np 2 ./a.out<br />
$ echo $?<br />
0<br />
$ mpicxx -cxx=icpc cl_solver_unsym_complex_c.cpp -lmkl_intel_thread -lmkl_core -lmkl_intel_lp64 -liomp5 -std=c++11 -DNSEGFAULT<br />
$ mpirun -np 1 ./a.out<br />
$ echo $?<br />
0<br />
$ mpirun -np 2 ./a.out<br />
$ echo $?<br />
0</p>
<p> </p>
Fri, 07 Nov 14 07:03:03 -0800qweasd q.535078Help! Negative output of IPARM (18) : Number nonzeros in factors.
https://software.intel.com/en-us/forums/topic/515127
<p>Hi,</p>
<p>Recently I solved a sparse<em> non-symmetric real matrix </em>equation with 2,785,003 unknowns,</p>
<p>The question is that I found <em><strong>Negative output of IPARM (18), </strong></em>which is the number of nonzeros in factors... , what's the meaning of this case?</p>
<p><strong><em>As a result, the solution produced is wrong.</em></strong></p>
<p>I wonder what's the problem, could anyone give me some information? Thanks a lot!</p>
<p> </p>
<p>//---------------------------------------------- output information when callong PARDISO -------------------------------------<br />
Linear Solver : MKL PARDISO is starting...</p>
<p> Reordering completed ...<br />
Number of non-zeros in factors =<strong><em> [-1.206401e+009] // This output is Negtive!</em></strong><br />
Number of factorization MFLOPS = [11120985]<br />
Factorization completed ...<br />
Solve completed ...</p>
<p> </p>
<p> </p>
Mon, 12 May 14 20:17:24 -0700yanpu z.515127Feast solver - internal memory error
https://software.intel.com/en-us/forums/topic/474787
<p>Hi all.</p>
<p>Calling dfeast_scsrgv, it returns with info=-1, an internal memory error.</p>
<p>The output is </p>
<blockquote><p>Extended Eigensolvers: Size subspace 10</p>
<p>Extended Eigensolvers: Resize subspace 0</p>
<p>Extended Eigensolvers: Error with Internal memory allocation</p>
</blockquote>
<p>I don't know if the error message is accurate, I am using the 64-bit version (11.1.0.103, Windows), and the matrices are pretty small, so it would need to be allocating a huge amount of memory to run out of address space. </p>
<p>Is it trying to create a temporary file - my C: drive is a SSD without much space left? Changing the program working directory to E: does not help, so is it creating a file in a specific location?</p>
<p>Is it a problem with the matrices - the Extended Eigensolver Functionality page in the help says A should be real symmetric, and B should be real symmetric positive definite - I don't have these, I have real symmetric positive definite, and B is real symmetric, about half of the rows in B (and corresponding columns) are zero, ie it could be reordered to <img src="http://latex.codecogs.com/png.latex?%5Cbegin%7Bbmatrix%7D%20B%20%26%200%5C%5C%200%20%26%200%20%5Cend%7Bbmatrix%7D" height="44" width="50" /> - this isn't positive definite, the submatrix isn't either, so should return an error of -3? But it seems to solve this for a different search range of eigenvalues (possibly the change makes some internally generated matrix much larger?)</p>
<p>Mathematically, the zeroes in the B matrix should give eigenvalues of +inf, and an eigenvector that is +1 for the entry corresponding to the matrix row, and zero for all other values, for each row/column that is zero. The other eigenvalues can be found by generating a modified A matrix - I dont know if the FEAST solver will do this internally, and solve the modified problem, or if these eigenvalues all get rejected because the search range is bounded, and the solver just works without B actually being positive definite.</p>
<p>The alternative is to solve the reverse problem, swapping A and B, and solving for 1/λ, but then the range of eigenvalues gets inverted, and I need to find the largest instead of the smallest. I might use ARPACK or something for this, FEAST doesnt seem particularly well suited for finding the largest.</p>
<p>Attached is a test program, and the a, b matrices. If changed to search a different range, test program will find 4 eigenvalues around 182.9, then another 4 around 228.</p>
<p>The benchmark I am testing against suggests I should find a couple of eigenvalues between 1 and 5, but I don't have access to Matlab to check my matrices directly, so the difference might be the matrices I am solving.</p>
<p> So, questions are:</p>
<ol>
<li>Is the out of memory error message correct, or is it a different problem?</li>
<li>Does B have to be positive definite?</li>
<li>Should I always be getting an error message with any matrix of this shape?</li>
<li>Are the values found accurate, or is B not being pos def causing FEAST to find a totally wrong solution?</li>
</ol>
<p>Thanks,</p>
<p>Alan</p>
<p> </p>
Wed, 25 Sep 13 11:08:35 -0700Alan Ritchie474787Pardiso unable to solve a symmetric matrix
https://software.intel.com/en-us/forums/topic/386684
<p>Hello MKL tech support,</p>
<p>I have a positive definite matrix and using Pardiso to solve the problem. I kept receiving the error -4 for the solver. It is interative refinement problem in phase 33.</p>
<p>My params for pardiso are:</p>
</p>
<p> iparm[7] = 15; <br /> iparm[10] = 0; <br /> iparm[12] = 0; <br /> iparm[0] = 1; <br /> iparm[1] = 0; <br /> iparm[3] = 0; <br /> iparm[4] = 0; <br /> iparm[5] = 0; <br /> iparm[6] = 0; <br /> iparm[9] = 13;<br /> <br /> iparm[11] = 0;<br /> iparm[13] = 0; </p>
<p> iparm[17] = -1; <br /> iparm[18] = 0; <br /> iparm[19] = 0; <br /> iparm[20] = 0; <br /> <br /> iparm[26] = 1; <br /> iparm[27] = 0; <br /> iparm[34] = 1; <br /> iparm[59] = 0;</p>
<p>I am trying to solve Ax = y but there was no error in the factorization phase until the phase 33.</p>
<p>Thanks,</p>
</p>
Tue, 09 Apr 13 14:35:07 -0700bryce155386684Pardiso produces a wrong answer
https://software.intel.com/en-us/forums/topic/373799
<p>Hi,</p>
<p>I have written a FORTRAN program that uses Pardiso to solve some ill-conditioned system of equations. 95% of the time, the code works just fine. But for some parameters, it calculates completely wrong solutions. I could trace this back to a specific matrix that isn't solved correctly by pardiso. The matrix has a condition number of about 10^8, so in double precision it should definitely be possible to calculate some approximate solution.</p>
<p>I have attached a very short example program that simply loads this specific matrix and the right-hand side from a file (the file is "unformatted", it's only called "txt" because this forum doesn't like "dat" extensions) and solves the system with Dgesv from the MKL version of LAPACK and with Pardiso. It then calculates the norm of the residual for both solutions. For the solution computed by Dgesv, you get about 2*10^-7, which is perfectly ok for my needs. For Pardiso, the norm is about 4*10^+6, so the "solution" is complete nonsense.</p>
<p>I'd be very grateful if you could check if I'm doing something wrong here or if there really is a bug in Pardiso.</p>
<p>We're still using MKL 10.3 update 9 in my company, so if this is a bug, but it has already been fixed, I'd be happy to know it so I can ask our administrator to install the latest version as soon as possible.</p>
<p>Best regards,</p>
<p>Lars</p>
Fri, 01 Mar 13 05:48:58 -0800Lars B.373799Is pardiso_64 back-substitution parallelized?
https://software.intel.com/en-us/forums/topic/346415
<p>Hello,</p>
<p>I've been running pardiso_64 on several problems. If I change MKL_NUM_THREADS I can quickly notice that the factorization elapsed time goes down.</p>
<p>However, the back-substitution doesn't seem to be affected by the number of threads. Given that I am solving, one right-hand-side at a time, many, many right-hand-side vectors, I would like the back-substitution to also be parallelized.</p>
<p>Do I need to do anything special to get speedup in this phase?</p>
<p>Thanks.</p>
Mon, 10 Dec 12 11:09:21 -0800Arthur M.346415Pardiso: error during symbolic factorization
https://software.intel.com/en-us/forums/topic/277605
<p>I'm using Pardiso to solve a very large sparse linear system and I get the following error message:ERROR during symbolic factorization: -2According to the description of the Pardiso errors, this corresponds to a problem with the memory allocation in Pardiso.I've doubled the memory available for the computation but I still get the same error message.Does anyone have any clues on how to get around this?Thanks!</p>
Wed, 27 Jun 12 00:43:05 -0700pf289277605