Concurrent GPU and NPU inference on IMX 8M Plus

[matthogg92]

Hello,

I would like to run 1 neural network on the VIP8000 Nano NPU and 1 network on the G7000UL GPU. Currently I am able to choose which accelerator to target through the USE_GPU_INFERENCE environment variable. Is it possible to override this environment variable and use both accelerators in an application running in the same environment?

Thanks :)

[sunshinemyson]

@matthogg92 ,

I suggest you ask your SoC provider to support this issue.

Thanks