Re: Guaranteeing processing speed...

Thomas Strohm (Thomas.Strohm@de.bosch.com)
Wed, 13 Oct 1999 14:36:32 +0200


Rogier Wolff wrote:
>
> You run the process with real-time priority. That means that the mpg
> player will get the CPU whenever it wants over other processes....
> This is fine as long as the program is fast enough to handle the
> datastream.

As long as you have just one (or just a few) processes with real-time
priority :-)

Thomas.

PS: Please CC me when replying...

> > Suppose you have a process that needs a certain amount of CPU power.
> > Think of
> > an MPEG player showing a movie. If you start other processes absorbing a
> > lot of
> > CPU power you may run into a problem with your MPEG player. Common
> > solutions
> > like
> >
> > - killing the other processes
> > - use a faster computer
> > - playing around with process priorities
> >
> > are not very satisfying. An interesting thing would be to have an
> > optional
> > parameter when fork()ing to request a certain CPU power (given in units
> > of
> > BogoMIPS or whatever). When the OS isn't able anymore to grant the
> > requested
> > amount of processing speed to the process, it could inform it by sending
> > a
> > signal. Then the process could decide upon what to do. The MPEG player,
> > e.g.,
> > could skip every second frame.
> >
> > Is this out-of-realm for Linux? Is it a typical problem for an RT-OS?
> >
> > Any thoughts about these problems?

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/