Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
07/09/10 11:00
Read: times


 
#177173 - re - Separate display processor
Responding to: ???'s previous message
Andy Neil said:
the idea of a co-operative system needs all tasks to complete before the scheduler runs again. This inevitably wastes processing time.
Does it?
The timer interrupt is asynchronous with anything else. How can one be certain that tasks are done before the interrupt fires unless tasks are shorter than the timer period?
If so, processor time is wasted.
I have been using a sparetime routine that runs at the end of the despatcher. This works well for eg., RS232.
It only needs to check that the remaining time is sufficient to send a character. This speeded up the comms tremendously.
Are you using polled comms, then?

No, this was done for the transmit part of the RS232. There is precious little time to send strings in between other tasks. Some tasks only poll a hardware input, so leave plenty of spare time that can be used.
Are we talking Coroutines here?

No, I do not think so.
The suggestion is for a background routine that will only be interrupted by the timer interrupt. Then it is returned to when the despatcher is done.
Coroutines are more complicated, I think.
Regards
Johan Smit



List of 23 messages in thread
TopicAuthorDate
co-operative scheduler            01/01/70 00:00      
   RE: This inevitably wastes processing time.            01/01/70 00:00      
      re - Separate display processor            01/01/70 00:00      
         I see (I think)            01/01/70 00:00      
            re - Separate display processor            01/01/70 00:00      
   Separate display processor?            01/01/70 00:00      
      re - Separate display processor            01/01/70 00:00      
   how can processing "waste processing time"            01/01/70 00:00      
      Answered here?            01/01/70 00:00      
      All multi-tasking with real-time requiremens suffer waste            01/01/70 00:00      
         re - Separate display processor            01/01/70 00:00      
            Parasitic newspapers?            01/01/70 00:00      
   I do not know your "Time Triggered Prioritised Cooperative S            01/01/70 00:00      
      re - Separate display processor            01/01/70 00:00      
         what does a time interval have to do with it?            01/01/70 00:00      
         TTPCS            01/01/70 00:00      
            re- TTPCS            01/01/70 00:00      
   No need for these canned "schedulers"            01/01/70 00:00      
      re - real time system            01/01/70 00:00      
         no, it does not            01/01/70 00:00      
            re- Scheduler            01/01/70 00:00      
      Timeslice timewaste            01/01/70 00:00      
         re - real time system            01/01/70 00:00      

Back to Subject List