Hi every body.
I have a question and i try to answer by myself but I think that not correct enough.
What is the reason why Kernel can not launch?
my answer is:
- check your Kernel define.
- check your threads in block to confirm that not over 512 threads per block.
- check the register per multiprocessor to confirm that enough for running.
- check shared memory per multiprocessor to confirm that enough for running.
- check your calling kernel.
- finally give the Error catching command to get information
“printf(“CUDA Error: %s\n”, cudaGetErrorString(cudaGetLastError()));”
In my program. i check all my answer [ CUDA Error is no error], even my kernel is very simple but it can not launch. This is the first time i got this situation. I tried to compile my old program and all of them work correctly.
can anybody give some advise.