in serverless GPU Is Delay Time also Charged or not??
Solution
^Yeah so there is:
Delay time which consists of:
(sitting in the queue + cold start (the machine starting up + maybe some setup code u might have outside the handler.py such as:
model = load(modelname) or whatever) + imports of libraries)
+ then the actual code running time.
as ashelyk said ur charged cold start to execution time, so basically only time that ur machine trying to run / is running
Delay time which consists of:
(sitting in the queue + cold start (the machine starting up + maybe some setup code u might have outside the handler.py such as:
model = load(modelname) or whatever) + imports of libraries)
+ then the actual code running time.
as ashelyk said ur charged cold start to execution time, so basically only time that ur machine trying to run / is running
