Integrate FDG Malloc

[delanyinspiron6400]

[delanyinspiron6400]

The issue may result from the fact, that it requires this warp-basis for allocations, so maybe we would have to pass in this warp state all the time.

// set namespace
using namespace FDG;

// get warp header
Warp* warp = Warp::start();

ptr = warp->alloc(5);

warp->end();

[delanyinspiron6400]

that does not seem to be the issue, but it seems that it always needs a full warp to function, on a thread-basis it does not work.

[delanyinspiron6400]

And even that does not work all the time, need to take a closer look unfortunately

[delanyinspiron6400]

It is placing some internal structures on CUDA*malloc'ed ptrs and they seem to be too small sometimes?
Debugging that is fun! :(