Jetson NANO 2GB devkit - how to communicate between CPU and GPU

I have a Jetson NANO 2GB devkit. I want to understand how I can communicate between the CPU and GPU. I do not want to use any IDE. I want to understand what the physical medium is and protocol, and how I would address the GPU.

e.g. in some systems I might use I2C or SPI. But I have no idea where to find this detail for the Jetson NANO 2GB devkit.

Any help appreciated.

Thanks

Mike

Moving into Jetson Nano forum.

Hi,
On Jetson platforms, we have NvBuffer APIs for processing data on CPU/GPU/other hardware engines. Functions are defined in

/usr/src/jetson_multimedia_api/include/nvbuf_utils.h

Generally we call NvBufferMemSyncForCpu()/NvBufferMemSyncForDevice() to make sure the process is done on CPU or GPU. There is no direct communication between CPU and GPU.

If you run a CUDA sample, you can see similar APIs for synchronization.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.