What is eval_time about DeepViewRT

[weilly_li]

Hi, 

    I have some comfuse about DeepViewRT.  What is eval_time ? Is really run time of model ?

    Code : "put_time, post_time, eval_time = modelclient.get_timing_info() "

    And.. What is put_time / post_time ? Is loding model time ? or Warm-up time ?

Weilly

[FelipeGarcia]

Hi Weilly,

Could you please provide what is the device you are working on? Also please let us know the code you are using for reference for this.

Best regards,

Felipe

[weilly_li]

Hi, @FelipeGarcia 

      My Platflorm is  NXP i.mx8mplus.

      Python Code :

         results = modelclient.run(inputs, outputs)

         put_time, post_time, eval_time = modelclient.get_timing_info() 

     So, Is DeepViewRT fast than Tensorflow Lite ?

Weilly

[Bio_TICFSL]

Hello Weily_li,

Regarding the 'get_timing_info' - please check {eIQ_Toolkit_Install|\docs\DeepViewRT_User_Manual.pdf -> A.2 ModelClient

Returns:

put_time: the timing information regarding how long it took to send the most recent rtm to the modelrunner application post_time: the total time to run the model (includes the post request and evaluation) eval_time: the time it took for the modelrunner application to run the model.

Can you share the model you used?

I assume the error you see appeared on the host PC. Any message shown on the board?

Please share the log from %appdata%Roaming\eIQ Portal\logs\

Regards