Re: First draft of "OROCOS II" proposal...

On söndag, jun 22, 2003, at 20:49 Europe/Stockholm, 
[email protected] wrote:

> On Sun, Jun 22, 2003 at 08:03:46PM +0200, 
> [email protected] wrote:
>> But it might be important in particular to consider how HARD real-time
>> graphics can be handled in a fashion similar to what is done for
>> example in QNX where the graphics is supposed to be real-time.
>
> May I ask what hard realtime graphics should be good for? I mean, the
> human looking at the screen is not hard realtime in terms of machine
> response times, so I don't see any application for such a beast.
>
>
In a queing based system the graphics can lack behind and you can end 
up with "delayed" or a "slow" system. You would want the graphics is 
throw away data if more recent data is in the pipe. You do not want the 
system designer to have to carefully consider "display" performance to 
make sure that everything can be drawn on the screen without a lock up, 
you what the graphics system to handle this automatically. If you post 
enough data to an X server it gives up life, which is not exactly the 
behaviour you are looking for. The alternative is that it takes up all 
CPU power which is equally poor in a real-time application.

Henrik