Interprocess communication with TX1

Autonomous Machines Jetson & Embedded Systems Jetson TX1
1

I am using the examples in 0_Simple - simpleIPC on tx1 and it hangs after commenting supporting only 64bit LinuxOS. I commented the line to move ahead and see what happens and it was stuck

[b]Launching kernels…

Process 1: Run kernel on GPU0, taking source data from and writing results to process 0, GPU0…
[/b]

I have an application where I need more analytics to process on an image after it passes through detectnet. I don’t want to free up the memory consumed by the image prior to passing it through detectnet and pass the pointer of the image to another process where the analytics can be done. I searched for IPC options and there is SimpleIPC which can achieve it but tx1 has a different architecture (aarch64 vs x86_64)

Anyone can guide me where to find simpleIPC examples for Tx1 specifically ? And also suggest how can I do the IPC with pointers on gpu memory ?

Thank you

2

Hi,

Please check our document for more information:
>> 3.2.8. Interprocess Communication:
https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#interprocess-communication

Thanks.

3

Thank you AastaLLL.

Upon reading, I realised tegra platforms donot support IPC API. Its only for x86_64 Linux versions. Thank you

1 Like