I just read about the limitation of global function parameter size to 256 byte on page 29 of the cuda programming guide.
what does that actually include?
I’ll giva an example:
__global__ void PointtIteration(int* pic, struct wbet _wbet,int max_iter)
int* pic is an array which has been allocated on the device by using the cudaMalloc function. -> its way bigger than 256 byte, so i estimate the limitation doesn’t affect the global device memory
which parameters are limited?
lets imagine, that the structure “wbet” contains an array of more than 256 bytes. will that work?
in case not, lets imagine, that i use 2 of these structures which have …lets say 192 bytes each. works?
im sick of getting a sprinkled screen, so i’ll ask before i try :D