While launced the kernel, error Checking API showing unspecified launch failure

Question

I wrote a CUDA program on Nvidia Tesla P100 with driver version 384.183. The code shows no error while not using any error check. But when the error check API is incorporate