Currently, I have a code that use C++ class.
I create one class : MyClass, which quite complex (contains template class, new type I defined and boost vector and matrix)
To allow multithreading I create a (boost) vector of objet of MyClass.
For instance to run 4 threads and a vector of 4 MyClass object, and run each thread on one object.
I would like to do the same with a GPU. Is it possible to define one object by CUDA cores ?
and how to compile the program with boost ? (I have Ubuntu 8.04)