My kernel have two arguments one is input other is output both of type char. In kernel I have taken two arrays:
Array_1 is calculated from Input and it has no problem and Array_2 is calculated from Array_1. And finally i have to assign value of Array_2 to output like this:
Here I got cudaErrorLaunchFailure but if I replace Array_2[ind].x…w on right hand side of = sign with numbers e.g. 45 etc then it executes fine.
So, Is any problem with excessing like Array_2[ind].x…w ?