Slow on using tensorflow in lambda after greengrass deployment

Question

I have tried deploying my lambda code in green grass deployment.   
The code seems to run slow when i use 'import tensorflow as tf' vs code without it. The code is the example helloworld counter code. I have yet to use any tensorflow commands  
It takes about 20-30 minutes before i there is a reply from the trigger.  
My greeengrass core device is a rpi  
  
Edited by: fishfish on Oct 20, 2019 4:59 PM

fishfish · Answer

it seems that the problem is solve by increasing the memory limit and timeout. not sure the reason though, could it be that the tensor flow library is bigger than 16MB?  
  
Edited by: fishfish on Oct 22, 2019 11:01 PM  
  
Edited by: fishfish on Oct 22, 2019 11:09 PM

Slow on using tensorflow in lambda after greengrass deployment

Contenuto pertinente