Hello.
My code calls the kernel, and the kernel checks operation needs.
and then in the kernel call the device function.
at that function, can I use the atomicOr (i.e., I wonder about using atomicOr in the device function)
because I learned operation is possible only in the kernel, but access is possible in device function
I read the guide.
I will try the using atomicOr() in the device function.
[url]https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#c-language-extensions[/url]
Yes, you can use all device-side intrinsics inside device functions as well. They are (at some nesting level) still called from a global kernel which is the only way to execute device code.
Thanks tera!