Hi.
Q1:
First in my simulation i have to choice CUfunction to load…
Each function have different used registers so i have to calculate max number of threads…
Is there any functionality to check number of used registers of currently loaded CUfunction during run time in host code?
Q2:
I’m using only *.x direction for parallelism…
uint idx = blockIdx.x * blockDim.x + threadIdx.x;
How to determine max value of threadIdx.x in current block/grid in device code?
I need to do something like:
//maxThreadX = ???
if(maxThreadX >= cntTriangs){
// Base store to shared memory
}else{
// Store to shared memory in loop
}
__syncthreads();
Thanks…
Regards BasY