This PR increases the maximum number of runs from 1000 to 8192. Besides simply changing the MAX_NUM_OF_RUNS define, in order for Cuda not to waste GPU memory it now, like OpenCL, allocates memory based on the actual settings of the number of runs and the population size instead of their max values.
This requires a little bit of additional testing on the Cuda side.
This PR increases the maximum number of runs from 1000 to 8192. Besides simply changing the MAX_NUM_OF_RUNS define, in order for Cuda not to waste GPU memory it now, like OpenCL, allocates memory based on the actual settings of the number of runs and the population size instead of their max values.
This requires a little bit of additional testing on the Cuda side.