We are currently using a Jetson Orin Nano and have run into an issue: the Orin Nano lacks hardware encoding support. In our current requirement, encoding is critical because we need to record and stream three cameras in parallel, which means a total of six encoders (three for recording and three for streaming). Software encoding cannot reliably handle this load.
We need to verify whether the Jetson Orin NX meets all of our requirements, which include:
-
Mic array: XMOS XK-VOICE-SQ664 4‑MIC Voice Development Kit
-
5G module
-
eSIM
-
GSM
-
UART
-
I2S
-
Wi‑Fi module
DeepStream pipeline requirements:
-
3 camera inputs
-
2 primary inference models (PeopleNet + TrafficCamNet)
-
3 secondary inference models
-
Recording and streaming: for 3 cameras in parallel, we need 6 encodes (recording + streaming for each camera).