Intel® Math Kernel Library

Pardiso allocation error

Hi, I tried to use Pardiso library to solve a sparse non-symmetrix complex matrix and I encountered an error message. This are details of matrix I tried to solve:

Linear System number of unknows = 113259
Matrix number of non-zero = 186565263
CSR storage format (about 3558.4 Mb in complex double precision)
RAM available = 128 Gb

This is the error message:

Pardiso Dll

I'm trying to use the pardiso function from my c# dot net application but I cannot find the Dll that it resides in. Does anyone know where I can find it?
I managed to create my own Dll by making a C++ Dll that imports and exports the pardiso function but I don't want to do that as it's more code to manage which is a pain for 32bit/64bit and dealing with new Mkl versions.

Also, are there any dot net examples for Mkl? It would be very useful.


Pardiso : memory release problem (MKL

Hi, I am trying to use the Pardiso solver for the nonstationary Navier-Stokes equations.
The only problem is that the system memory is increasing constantly.

I tried MKL_FreeBuffers(), but it was not working.
I am curious to know if the problem could be fixed.
Is anyone who use Pardiso solver to the dynamic PDE problems?
Any comment would be appreciated.

Large size of fortran executable when compiling with intel MKL

Sorry if my thread appears twice, but I was not logged in when I sent it the first time, and as I don't know if that is an issue or not, I am resending it after logging in.

I am running on linux redhat enterprise v.7. Previously, I was compiling my fortran 90 programs with the lahey fujitsu compiler and statically linking to the lapack and blas libraries that came with the lahey version. The size of the executable was about 2 MB.

Find only some eigenvectors with a banded matrix?

Can this be done with the bisection followed by inverse iteration method?

I notice there don't seem to be any banded routines for finding only some e-vectors.

I tried the driver routine dsbevx, but it apparently must be using QR or divide-and-conquer because it requires an eigenvector matrix big enough to hold all eigenvectors, even if it's told to return only a few.It also takes the same amount of time to solve a 1000 DOF system for 4 e-vectors as it does to solve the same system for 200 e-vectors.

porting to 64bits: where is the "right" solver / mkl library?


i'm in the process of porting my vc++ 2008 application to 64bit. in my "personal" interface for pardiso, i used to link in pardiso by using the following:

#pragma comment(lib, "mkl_c.lib")
#pragma comment(lib, "libguide.lib")

this would essentially also pull mkl_ia32.lib, mkl_lapck.lib and mkl_solver.lib. i don't find the 64bit replacements for these libraries. (mostly, because i don't find anything regarding the library naming conventions in the docs.)

Assine o Intel® Math Kernel Library