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

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
12/26/07 21:12
Read: times


 		 #148745 - Worst case, instead of average
Responding to: ???'s previous message
Firstly, my comment you referred discusses about Oliver's single UART byte (single ADC result) buffering. Single byte buffering has no difference with the direct polling on the main loop. These two method have the same 'life time' of the variables. On the single byte buffering, rather, the interrupt latency reduces the effective execution time of the firmware. Then, I recommended multi-byte buffering (or process) in the ISR.

Secondly, the timing should be always designed according to the worst case. There is no room to consider on average. Circular buffers extend the minumum character separation time to those multiplied by the buffer size. This span is compared with the maximum main loop time.

Tsuneo

List of 34 messages in thread
TopicAuthorDate
Having problem on using interrupt            01/01/70 00:00      
   Answers            01/01/70 00:00      
      not so fast, Cristoph and to Ragu            01/01/70 00:00      
         Serial comm is slow.            01/01/70 00:00      
            if you read the OP            01/01/70 00:00      
               good interupts            01/01/70 00:00      
                  AAAAAARGH            01/01/70 00:00      
                     not AAAAAAAARGH            01/01/70 00:00      
                        OH YES            01/01/70 00:00      
                           do You ever avoid "non-debuggable"?            01/01/70 00:00      
                              always            01/01/70 00:00      
                                 nice            01/01/70 00:00      
                                    a trick for "long ISRs" better than a RTOS            01/01/70 00:00      
         Buffered RX            01/01/70 00:00      
            not much difference            01/01/70 00:00      
               You are right            01/01/70 00:00      
   Sucess with interrupts            01/01/70 00:00      
      Isn't it too short?            01/01/70 00:00      
         the reason for circular buffers            01/01/70 00:00      
            Worst case, instead of average            01/01/70 00:00      
               absolutely, just worst case is not as bad            01/01/70 00:00      
                  single, double and ring buffer            01/01/70 00:00      
                     single, double and ring buffer app dependent            01/01/70 00:00      
         and some more reasons to use (circular) buffers            01/01/70 00:00      
            buffer overflow error            01/01/70 00:00      
               Buffer overflow handling            01/01/70 00:00      
                  Buffering options            01/01/70 00:00      
                     Beauty contest !            01/01/70 00:00      
                        flexible is the best?            01/01/70 00:00      
                        Simplicity is always better            01/01/70 00:00      
                           simple is another word for            01/01/70 00:00      
                              Where is the borderline to stumble on            01/01/70 00:00      
                                 Copper Oxide            01/01/70 00:00      
               please exclude me from the 'we'            01/01/70 00:00      

Back to Subject List