On Oct 17, 5:47 pm, Andy <
andy.py...@gmail.com> wrote:
> Hi is there a way to use CuPP to implement a tree data structure with
> dynamic memory allocation?
Are you talking about dynamic memory allocation on the host or on the
device? I guess the later. This is not included, but you can easily
simulate such a behavior, by allocating a big memory block of device
memory on the host and than hand chunks of this block out at the
runtime of your kernel.
In case you have some more time to wait for this feature, dynamic
memory allocation is named in the Fermi docs, so it should be
available in an upcomming CUDA version.
-Jens