[Xcb] profiling and performance

[Vincent Torri]

On Tue, 2 May 2006, Donnie Berkholz wrote:

> Vincent Torri wrote:
> > there are other xcb functions that take more time :
> >
> > write_block
> > _xcb_in_read
> > read_packet
> > etc...
>
> 1038      0.1930  libXCB.so.0.0.0          write_block
> 574       0.1067  libXCB.so.0.0.0          _xcb_in_read
> 571       0.1062  libXCB.so.0.0.0          read_packet
> 514       0.0956  libXCB.so.0.0.0          get_event
>
> The top time-taker in xcb is 0.2% of the time spent in your app. Do you
> honestly think that's worth optimizing? You might get down to 0.15% or
> 0.1%, but chances are the real-world difference will be practically
> nonexistent.

I don't know. But the Xlib engine is faster. There must be a reason, no ?
Jamey told me to tell him if there is something else than the send request
function that takes time. So I tell him.

Also, optimizing is always a good thing. Even if it's some few percent.
Old and slow computers exist. This percentage can be a lot more on these
ones.

Vincent