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

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
06/20/06 21:51
Read: times


 		 #118653 - DMA, logic and stuff...
Responding to: ???'s previous message
Richard Erlacher said:
I won't argue the matter of DMA as a concept, at this time, BUT, from the standpoint of implementation, it requires, not only that there be a good deal of additional logic, BUT, that there be a good deal of additional logic interposed between the CPU core and all its resources, in order to give the DMAC access to the same pins (both internal and external) that the CPU core uses to access those same resources. [... etc. I think I get your point by now JW]

Don't get me wrong. If you want a softcore '51 with zero or minimum SFR, peripherals, anything, have it. Enjoy. I am only making a couple of suggestions and hints - they may be stupid and completely wrong, of course, you know me by now, I chatter nonsense from time to time... :-)

But, the point is, if you want a hyperfast core, you WON'T build a bottleneck multiplexing code and data memory bus. Hence, while the DMA controller or anything similar happily reduces the performance of DATA access (which is already bad, having only 1 16-bit pointer which takes ages to load/modify), while the CODE memory interface stays untouched running as fast as it can...

0.02SKK (practically zero) :-)))

JW

List of 95 messages in thread
TopicAuthorDate
How badly would this hurt ...            01/01/70 00:00      
   Depends            01/01/70 00:00      
      dam'd americans            01/01/70 00:00      
   not a new variant, exactly            01/01/70 00:00      
      Use the Power of the 8052            01/01/70 00:00      
         which history?            01/01/70 00:00      
         That's my first choice ... however ...            01/01/70 00:00      
      DMA            01/01/70 00:00      
         No such plan is in the works.            01/01/70 00:00      
            DMA!            01/01/70 00:00      
               OK, but how much will you give up?            01/01/70 00:00      
                  DMA, logic and stuff...            01/01/70 00:00      
                     I think you have a different concept            01/01/70 00:00      
                        I don't understand... but...            01/01/70 00:00      
                           I'm not sure where this is headed            01/01/70 00:00      
      Cool            01/01/70 00:00      
         TI writable            01/01/70 00:00      
         if this is it, it is solved, if not rest            01/01/70 00:00      
   major confusion            01/01/70 00:00      
      Aye ... there's the rub...            01/01/70 00:00      
         a loud and resounding NO WAY            01/01/70 00:00      
            too late to edit            01/01/70 00:00      
               That's not what I want ... right now ...            01/01/70 00:00      
                  it DOES matter            01/01/70 00:00      
                     not in this case            01/01/70 00:00      
                        XDATA vs DATA/IDATA            01/01/70 00:00      
                           it's a work in progress, and, like many            01/01/70 00:00      
                        We seem to have come a long way            01/01/70 00:00      
                           yes, it's a mess ...            01/01/70 00:00      
                              if this is a survey            01/01/70 00:00      
                                 I'd have to agree            01/01/70 00:00      
                                    you can have your cake and eat it too.            01/01/70 00:00      
                        sure it does            01/01/70 00:00      
                           depends on memory width, doesn't it?            01/01/70 00:00      
                              go ahead, but PLEASE do not refer to the            01/01/70 00:00      
                                 I don't know what one should call it            01/01/70 00:00      
                                    I do, when it has benefits            01/01/70 00:00      
                                       I looked at them years ago            01/01/70 00:00      
                                          you are missing 'enhancements'            01/01/70 00:00      
                                             well if it uses the SFR's            01/01/70 00:00      
                                                as an example of an enhancement            01/01/70 00:00      
                                                   This is field-programmable logic.            01/01/70 00:00      
                                                      and have one more development step/uncer            01/01/70 00:00      
                                                         That's the whole point ...            01/01/70 00:00      
                           back up a bit ...            01/01/70 00:00      
                              AR7 is the annotation used by the better            01/01/70 00:00      
               Actually, that's not rational            01/01/70 00:00      
   It wouldn't bug me at all            01/01/70 00:00      
      not good enough            01/01/70 00:00      
         someone didn't read the datasheet first            01/01/70 00:00      
            that is impossible, ridiculous as well            01/01/70 00:00      
   not a thought but the fact            01/01/70 00:00      
      How to make a simple process ugly            01/01/70 00:00      
         new core            01/01/70 00:00      
            Atmel says            01/01/70 00:00      
               of course, they do            01/01/70 00:00      
               Why not just switch to another mfg?            01/01/70 00:00      
                  availability            01/01/70 00:00      
                     too bad ... how about a 68HCsomething?            01/01/70 00:00      
                     no problem            01/01/70 00:00      
                        Today Russia is            01/01/70 00:00      
      Be careful with ATMEL ...            01/01/70 00:00      
         What's with the French bashing?            01/01/70 00:00      
            French bashing            01/01/70 00:00      
               nothing wrong with french bashing            01/01/70 00:00      
                  There's nothing pressing            01/01/70 00:00      
                  You guys should be ashamed of yourselves            01/01/70 00:00      
                     Oh, LIGHTEN UP!            01/01/70 00:00      
            Probably this            01/01/70 00:00      
               Yes, that played a significant part            01/01/70 00:00      
                  I agree half the way            01/01/70 00:00      
                     Well, I'm P*ssed off!            01/01/70 00:00      
                        of course not            01/01/70 00:00      
   there are ways            01/01/70 00:00      
      Where does this fit?            01/01/70 00:00      
         with code as above you can feed a 460k U            01/01/70 00:00      
            I don't doubt that for a moment, but ...            01/01/70 00:00      
               already answered            01/01/70 00:00      
                  they're cleared independently, right?            01/01/70 00:00      
                  you get to the ISR, how do you know why?            01/01/70 00:00      
                     isn't that TI and RI            01/01/70 00:00      
                        I'm puzzled too            01/01/70 00:00      
                     you have two flags            01/01/70 00:00      
                        and so what?            01/01/70 00:00      
                           That would cause a problem.            01/01/70 00:00      
                              Richard, sleep on it            01/01/70 00:00      
                                 It's exactly how Moto does it            01/01/70 00:00      
                                 Maybe not ...            01/01/70 00:00      
                                    JBC            01/01/70 00:00      
                                       it's not clear yet            01/01/70 00:00      
                                    What is the difference, they take the sa            01/01/70 00:00      
                                 to Richard            01/01/70 00:00      
                                    Thanks, I'll give it a look            01/01/70 00:00      
   you get to the ISR, how do you know why?            01/01/70 00:00      
      don't RI and TI tell you?            01/01/70 00:00      

Back to Subject List