How to deploy onto NPU of Sentinel?

[gerald.wang]

I reviewed the posts. But I didn't find discussions on deployment onto NPU of Sentinel. Would you please advise on this? Thanks.

[tom]

@gerald-wang https://docs.modalai.com/voxl-tflite-server/#deep-learning-on-voxl-2s-gpu-and-npu

[gerald.wang]

@tom Thanks Tom. I already checked the link. I understand that for gpu delegate "gpu" needs to be specified in /etc/modalai/voxl-tflite-server.conf. My question is what value needs to be used in this configuration file to trigger npu deployment. According to my exploration, specifying "npu" is the same as specifying "gpu".

cpu monitoring reports there are 8 cpus for sentinel. Which of the 8 cpus are opus?

Looking forward to hearing from you.