I’m developing a watch face. It has a rendering pipeline that’s somewhat elaborate so I’m keeping an eye on the performance.
Since I don’t have a physical watch (it’s a Pebble Round 2 on pre-order…) I’m developing in the emulator. However I understand that the emulator is likely somewhat faster than the actual watch.
I don’t want to be in a position where my watch face works great on the emulator, but is too slow to be practical in real life (or takes up too much battery).
Is there a general ballpark figure of how much faster the emulator is? For example, if this was Linux, I’d just cat /proc/cpuinfo and look for the BogoMips. Which is not accurate at all, but does give me a high-level picture, which is all I need right now.
Or, is there a way to run the emulator at actual hardware speed, like a retro console emulator?
Thece’s not, unfortunately. That would be really nice.
If it’s lagging even a little in the emulator it will definitely lag on real hardware. But the amount of slowdown really depends on what’s hogging the CPU so it’s a little hard to tell for sure
Is there a rule of thumb for what kind of performance or “energy budget” is acceptable?
Like, for a watch face that updates once per minute: I guess 5 milliseconds is entirely OK, and 5 seconds is going to kill the battery… is 50 milliseconds too much? 500 milliseconds? (assuming it’s maxing out the CPU for the duration)