Re: About render speed

[Weng Xuetian <wengxt gmail com>]

On Sun, May 20, 2012 at 11:51 PM, Rui Tiago Cação Matos
<tiagomatos gmail com> wrote:
> On 20 May 2012 15:59, Weng Xuetian <wengxt gmail com> wrote:
>> I'm implementing an extension that may need to update a window very
>> frequenctly, say maybe 10 times per second. (Please don't ask about reducing
>> the frequency, this is based on an very rapid user input and it need to be
>> updated in time to give user correct reaction.)
>>
>> I'm using 10 St.Text and all of them will be updated, but I found it could not
>> be updated in time (very obvious delay, say second level).
>
> I don't have a ready answer but I'd like to point that there is a
> lower bound on the achievable latency with the current architecture
> just due to the various context switches that must occur for you to
> see stuff on the screen after the user presses a key:
>
> X server > App > DBus daemon > Input Method daemon > DBus daemon >
> gnome-shell > X server
>
> So that's 6 context switches per event. Now, there probably is some
> low hanging fruit that could greatly improve things even with the
> current architecture but there will be a lower bound which might still
> be noticeable.
>
> Rui

Thanks, but the problem is, when I'm using only one St.Text and,
concat the string I want (yeah, kinds of workaround), there is no much
obvious latency, so I wonder how can I do it better since I want easy
separate event handle on each of the text.

The difference is simple, 1 St.Text and 10 St.Text, but I don't expect
there is so much latency that people can notice.