For example, if microphone, touchscreen, mouse movements, and camera are read to CPU and processed per sound sample by arbitrary code (which is not the bottleneck) then sent to speakers, how far behind do the speakers lag behind those inputs? Common phones have lots of input lag, like when dragging to scroll a browser you can see it scrolling as how you moved a fraction of a second ago, and I’ve tested this with custom apps to know its not intentionally delayed. For example, the same .jar file (such as using my jsoundcard) has much lower lag between microphone and speakers on linux than on windows. What is the input lag of all the inputs and outputs, and paths between them, on librem5?
Since we are still designing the hardware of the Librem5 we honestly do not know yet and since we will be using parts in the final phone that we will probably not use in the development kit we also can not evaluate this any time soon.
@BenRayfield it’s probably good that Nicole does not set any expectations.
However, one has to keep in mind that the lagginess of Android is mostly due to the layers of abstraction and the memory bloat + ungenerous RAM sizes.
I assume this ineffectiveness is a major reason for Google to work on Fuchsia.
The hardware should not be a problem. Try with a Raspberry Pi. Even back in the nineties one could have non-laggy applications on a 0.1 GHz PC.
So, there’s a good chance if you find the Librem 5 to be too laggy next January, it can be tweaked a lot on the kernel/driver side, toolkit side and the actual apps.
[EDIT:] Pondering on that: Maybe this could become one of the big selling-points to “regular”, non-tinfoil-hat users: show to a friend how you tap a notification and instantly Matrix shows the new message. You tap the URL in Matrix and before you know it, the browser is open, playing a video. That would be rad
There is a reason I turn off animations in Android . . .
You can either make something a design constraint and vary the hardware andOr OS to solve it, or solve whats really important to you and let it happen by chance. Why isnt specific amounts of low lag a design constraint?