I currently playing with cudaLaunchKernel function.
What i want to do is check the argument of the kernel but since arguments are passed by double-pointer, I can’t access them.
I found this document, CUDA Driver API :: CUDA Toolkit Documentation cuLaunchKernel, And it says
“The number of kernel parameters and their offsets and sizes do not need to be specified as that information is retrieved directly from the kernel’s image.”
I’m having trouble understanding this sentence. what does the kernel’s image mean? and Can i retrieve that information too?
Is there any way that I can get the number of arguments or any other way to check the arguments?