TensorFlow model not using swap memory even when using cpu

Robotics & Edge Computing Jetson Systems Jetson Nano
1

I added swap memory for my Jetson Nano because the 4GB was not enough. Because GPU memory is not shared with the swap memory, I instead tried to do inferencing with cpu and did:

os.environ['CUDA_VISIBLE_DEVICES'] = '-1'

However, only 25% of swap memory is used and then out of memory error. Does this mean I was not using the CPU all along?

3

hi @ju81153

Can you please refer this topic, hope this will help you Swap memory is not being used by the code

Thank you

4

Hi,

Could you share the output log of TensorFlow with us first?
Thanks.