We want to run Nvidia’s Redtail algorithm on the Jetson nano with 2 GB of ram. The goal is to make a drone that navigates through obstacles. Will 2 GB be enough? as the 4GB model is out of stock everywhere in my country, If not, can I use an SD Card as a swap memory and make it work somehow? Thanks
Hi,
The sample is released several years ago so you may need some update to make it work.
Usually, we need ~1GiB memory for loading CUDA, cuDNN and TensorRT library.
So it will be very limited to deploy a DNN model on Nano 2GB.
More, the swap memory is not a GPU-accessible memory.
It won’t increase the capability of DNN inference.
We also have another devices that equipped with more memory.
You can find more details in the below page:
Thanks.
1 Like
Thank you so much
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.