Unified and pinned memory vs in jetson NX

hi guys.
I have used bellow script in jetson xavier NX to gray my rgb image. how can I avoid cudamemcpy in my script by using unified or pinned memory to reduce time of execution?
does time of bellow script reduces by using unified or pinned memory or no?
I tested Pinned memory and used cudaevent to get time of execution but it made the program 3 times slower.
what is the solution
script :


Depends on the usage (read/write requirement), the recommended memory type will be different.
You can find the detailed memory type behavior in the below document:

If you want to use OpenCV → CUDA, below is a good sample for your reference:


This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.