several dynamic arrays is shared memory

Accelerated Computing CUDA CUDA Programming and Performance
1

I wonder how i can use several dynamically allocated arrays in the shared memory.
After some googling i found the following solution

extern shared int s_d1[];
shared int *s_d2 = &s_d1[num_el_in_s_d1];

where one has to pass the total size of the shared memory to be allocated as a third argument in the kernel.

However if I perform this then I get the following error message:

error: initializer not allowed for shared variable

Is there a standard way to do this? How does this change if one includes shared arrays of different variable types?

2

Remove [font=“Courier New”]shared[/font] from the second declaration, as that would place the pointer itself in shared memory. The compiler is smart enough to detect that the memory pointed to is shared memory.

3

Terrific! Thanks!