I wanted to know if I could stop Eigen from new memory automatically and if I would get a speedup by doing so. I've attached a code snippet which is somewhat characteristic of what my program does.

#define EIGEN_RUNTIME_NO_MALLOC


#include <Eigen/Dense>


using namespace std;

using namespace Eigen;

#define UNCONST(t,c,uc) Eigen::MatrixBase<t> &uc = const_cast<Eigen::MatrixBase<t>&>(c);

template <typename Derived>
void multiply (const MatrixBase<Derived> &A, MatrixBase<Derived> const &const_C){
Matrix<double, Dynamic, Dynamic>B;
B.setOnes(500,500);

UNCONST(Derived,const_C,C);

Eigen::internal::set_is_malloc_allowed(false);
C = A*B;
Eigen::internal::set_is_malloc_allowed(true);
}

int main(int argc, char** argv)
{
Matrix<double,Dynamic,Dynamic>A,C;
A.setOnes(500,500);
C.setOnes(500,500);
multiply(A,C);

}
I compile this code in debug mode and when I execute it, I get the following error:

Assertion failed: (is_malloc_allowed() && "heap allocation is forbidden (EIGEN_RUNTIME_NO_MALLOC is defined and g_is_malloc_allowed is false)"), function check_that_malloc_is_allowed, file ../3rdparty/Eigen/src/Core/util/Memory.h, line 189.

Is there a way I can switch off this memory allocation ? I'm careful of making sure that matrices on the left hand side of an assignment have the correct dimensions. Would there be speedups in the code if I stopped these malloc calls ? I deal with large matrices and I use functions like multiply() repeatedly.

Thanks a lot for your time!

Best,

Ashish

yes:

C.noalias() = A*B;

see: http://eigen.tuxfamily.org/dox/TopicWri ... ssion.html