topic Re: pdgetrf in mixed language in Intel® oneAPI Math Kernel Library

pdgetrf in mixed language

phaser75 — Thu, 24 Sep 2009 09:20:50 GMT

I exercise LAPACK in c language. I want to factorize a matrix(2X2) but it doesn't work.

I don't have any idea what is the problem.

The simple code is like this. The error message says there is a problem in 4th parameter(lda).

#include
#include
#include "mkl.h"

#define NRANSI
#include "NRUTIL.h"

#define PI (3.141592653589793)

int main(int argc, char* argv[])
{
int m, n, lda, info;
double **A, d;
int *ipiv;

A = dmatrix(1,2,1,2);
ipiv = ivector(1,2);

m = 2;
n = 2;
lda = 2;
A[1][1] = 4.0;
A[1][2] = 1.0;
A[2][1] = 3.0;
A[2][2] = -1.0;

printf("This is TEST!!!\n");

dgetrf(&m, &n, &A[1][1], &lda, &ipiv[1], &info);
printf("info = %d\n",info);

free_dmatrix(A,1,2,1,2);
free_ivector(ipiv,1,2);

return 1;
}

Re: pdgetrf in mixed language

Gennady_F_Intel — Thu, 24 Sep 2009 10:44:22 GMT

Quoting - phaser75

I exercise LAPACK in c language. I want to factorize a matrix(2X2) but it doesn't work.
I don't have any idea what is the problem.

The simple code is like this. The error message says there is a problem in 4th parameter(lda).

#include
#include
#include "mkl.h"

#define NRANSI
#include "NRUTIL.h"

#define PI (3.141592653589793)

int main(int argc, char* argv[])
{
int m, n, lda, info;
double **A, d;
int *ipiv;

A = dmatrix(1,2,1,2);
ipiv = ivector(1,2);

m = 2;
n = 2;
lda = 2;
A[1][1] = 4.0;
A[1][2] = 1.0;
A[2][1] = 3.0;
A[2][2] = -1.0;

printf("This is TEST!!!n");

dgetrf(&m, &n, &A[1][1], &lda, &ipiv[1], &info);
printf("info = %dn",info);

free_dmatrix(A,1,2,1,2);
free_ivector(ipiv,1,2);

return 1;
}

please try in that way:

A[0][0] = 4.0;
A[0][1] = 1.0;
A[1][0] = 3.0;
A[1][1] = -1.0;

dgetrf(&m, &n, &A[0][0], &lda, &ipiv[1], &info);

it should work.

--Gennady

Re: pdgetrf in mixed language

phaser75 — Thu, 24 Sep 2009 10:54:05 GMT

Quoting - phaser75

I still have the error.

The message is "MKL ERROR : Parameter 4 was incorrect on entry to DGETRF
info = -4".

Re: pdgetrf in mixed language

Gennady_F_Intel — Thu, 24 Sep 2009 10:55:42 GMT

Quoting - Gennady Fedorov (Intel)

Quoting - phaser75

I exercise LAPACK in c language. I want to factorize a matrix(2X2) but it doesn't work.
I don't have any idea what is the problem.

The simple code is like this. The error message says there is a problem in 4th parameter(lda).

#include
#include
#include "mkl.h"

#define NRANSI
#include "NRUTIL.h"

#define PI (3.141592653589793)

int main(int argc, char* argv[])
{
int m, n, lda, info;
double **A, d;
int *ipiv;

A = dmatrix(1,2,1,2);
ipiv = ivector(1,2);

m = 2;
n = 2;
lda = 2;
A[1][1] = 4.0;
A[1][2] = 1.0;
A[2][1] = 3.0;
A[2][2] = -1.0;

printf("This is TEST!!!n");

dgetrf(&m, &n, &A[1][1], &lda, &ipiv[1], &info);
printf("info = %dn",info);

free_dmatrix(A,1,2,1,2);
free_ivector(ipiv,1,2);

return 1;
}

please try in that way:

A[0][0] = 4.0;
A[0][1] = 1.0;
A[1][0] = 3.0;
A[1][1] = -1.0;

dgetrf(&m, &n, &A[0][0], &lda, &ipiv[1], &info);

it should work.

--Gennady

the same with &ipiv[0] not &ipiv[1]

Re: pdgetrf in mixed language

phaser75 — Fri, 25 Sep 2009 01:16:02 GMT

Quoting - Gennady Fedorov (Intel)

the same with &ipiv[0] not &ipiv[1]

Thank you for replying but it still doesn't work. Error message is same.

Re: pdgetrf in mixed language

ArturGuzik — Fri, 25 Sep 2009 02:28:48 GMT

Quoting - phaser75

Thank you for replying but it still doesn't work. Error message is same.

Hi,

check this thread. You have double indexed matrices here.

for statically allocated matrices example of using LAPACK is here.

A.

Re: pdgetrf in mixed language

phaser75 — Fri, 25 Sep 2009 04:07:18 GMT

Quoting - ArturGuzik

Hi,

check this thread. You have double indexed matrices here.

for statically allocated matrices example of using LAPACK is here.

A.

Thank you. But it is the same.

I tried static allocation and one dimensional array as well. But the same error occurred.

What's the problem?

phaser.

Re: pdgetrf in mixed language

ArturGuzik — Fri, 25 Sep 2009 06:41:16 GMT

Quoting - phaser75

Thank you. But it is the same.
I tried static allocation and one dimensional array as well. But the same error occurred.

What's the problem?

phaser.

OK then. This is working for me (static for simplicity).

[cpp]//
#include 
#include "mkl_lapack.h"
#define MAX 2

int main(){
   
  // data
   int n, m;
   int nrhs = 1;
   double A[MAX][MAX];
   double b[1][MAX];
   int lda = MAX;
   int ldb = MAX;
   int ipiv[MAX];
   int info;
   // Other values
   int i,j;

	// dimension(s)
	m=2;
	n=2;

	// value(s)
	A[0][0] = 2.0;
	A[0][1] = 0.0;
	A[1][0] = 0.0;
	A[1][1] = -2.0;
	b[0][0]=1.0;
	b[0][1]=2.0;

	// factorize
	dgetrf(&m, &n, &A[0][0], &lda, &ipiv[0], &info);
	// solve
	dgetrs("N", &n, &nrhs, &A[0][0], &lda, &ipiv[0], &b[0][0], &ldb, &info);

   // Check for success
   if(info == 0)
   {
      // Write the answer
      std::cout << "The answer isn";
      for(i = 0; i < n; i++)
        std::cout << "b[" << i << "]t" << b[0] << "n";
   }
   else
   {
      // Write an error message
      std::cerr << "dgetrf returned error " << info << "n";
   }

   return info;
}
[/cpp]

Re: pdgetrf in mixed language

redleaf — Thu, 11 Aug 2011 22:58:41 GMT

I've got the same problem with using the MKL library.

I've just found out that the problem lies in the parameters' data type.

Actually there's some incompatibility between Int32 and Int64.

Declared as MKL_INT, variables were interpreted as Int32 at compile time.

Nevertheless it was linked with the mkl_intel_ilp64, hence when I pass the address of an "int" instead of a "long int", the function might get a zero value. In this case the matrix leading. It leads to this error.

The problem is solved by simply changing the type of lda, m and n to long int.