Input latency

Kevin Gunn kevin.gunn at canonical.com
Mon Dec 16 23:30:40 UTC 2013


hey Daniel - thanks for the numbers, however the last statement of having a
core 100% loaded with mir_demo_client_egltriangle doesn't seem reasonable.
running on N4 shows ~20% load on cpu at most.

br,kg


On Mon, Dec 16, 2013 at 2:40 AM, Daniel van Vugt <
daniel.van.vugt at canonical.com> wrote:

> Nexus10 (3.4.0-4-manta)
> Direct 1.0ms with sporadic spikes to 74ms, sometimes 400ms
> Nested 1.2ms with sporadic spikes to 100ms, sometimes 1000ms
>
> I'm not convinced the problem is specifically Android. It could just be a
> common issue visible on the slowest hardware.
>
> On the Nexus10, just running a single server and client (egltriangle)
> occupies almost 100% of one of the cores. So stressing the CPU doesn't
> leave much time for being responsive.
>
>
>
> On 13/12/13 17:31, Daniel van Vugt wrote:
>
>> Here are some fun numbers I've collected about the latency between input
>> events sent from the top-level Mir server to a client. All in
>> milliseconds...
>>
>> Desktop (3.12.0-7-generic)
>> Direct 0.8ms
>> Nested 1.3ms
>>
>> Desktop (3.11.0-11-lowlatency)
>> Direct 1.0ms
>> Nested 1.7ms
>>
>> Nexus4 (3.4.0-3-mako)
>> Direct 0.9ms
>> Nested 1.5ms with high variance; frequent spikes to 73ms. Sometimes 700ms.
>>
>> The above numbers are for motion events. Key events appear to be twice
>> as fast (half the latency). This suggests we could yet optimize the
>> processing of motion events to perform better like key events do.
>>
>> In summary, we're looking pretty good. Only the Nexus has potential
>> usability issues, and even then, only with nested servers.
>>
>> - Daniel
>>
>>
> --
> Mir-devel mailing list
> Mir-devel at lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/mir-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/mir-devel/attachments/20131216/984be52b/attachment.html>


More information about the Mir-devel mailing list