I am only going by what I have seen. According to Lewis Heidrick, "runtime over 100 is suspect".
To further quote Lewis: (formatting edited by me)
I don't know what the real answer here is, but a few hundred millseconds at most would seem to be normal. If you think about how long it should take to sense something, contact the hub, and to process that and do an action, you are talking about maybe 500ms (1/2 second) or so. I would say 1000ms or more would be pretty ridiculous with this processor and memory, UNLESS you are involving the cloud or doing some fairly intense rule logic processing.
I do see very heavy and long processing with Echo Speaks on my hub, for example. It is the worst of any app I have. But I can see why with the heavy cloud reliance, and how large the app is, and how many app and driver codes it takes to make this app.
Even then, it doesn't seem to affect my hub, maybe because this is the only app that is so memory intensive? Everything else is a fraction of what I see compared to the Echo device and Echo app shown below.
device id 442 runcount 28 total runtime 47108 average run time 1682.4285714286
app id 309 runcount 140 total runtime 49531 average run time 353.7928571429
Other apps, like a motion lighting and it's sensor are more something like this: (stats taken a few seconds apart so they don't match how many times they have run)
device id 801 runcount 5 total runtime 46 average run time 9.2
app id 1127 runcount 2 total runtime 218 average run time 109
