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

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
04/15/09 18:59
Read: times


 
#164629 - No, it should NOT be just that ...
Responding to: ???'s previous message
Per Westermark said:
Richard said:
When the ARM with 32 KB of code space + 8 KB of data space can support the same task at the same cost with the same effective performance as the 805x, the 805x will truly be "circling the bowl." In theory, the ARM will be more development-friendly.

That shoud be "When the ARM with whatever amount of code and RAM space can support the same task at the same cost..."

The actual size of flash or RAM is irrelevant as long as there is enough, and the price is acceptable. It is a well-known fact that the cost of a microcontroller do not follow the size of the memory regions. More memory is normally followed by more peripherals and gives a higher price within the same family from the same manufacturer. But as soon as you move to another family from the same or other manufacturer, the price correlation fails. I'm currently looking at a move from one NXP ARM family to another. I gain more RAM and more peripherals, but since I move to a newer architecture I'll save money at the same time.


When making a comparison based on consumed resources, the relative performance has to be considered as well. If I can use an ARM at the same cost but have to tolerate performance that fails to meet my requirements, I'm unable to "get by." If an ARM chip with 64TB of memory on-chip costs the same as an 805x with only 64KB (the most it's normally likely to have) and both of them cost within 1% (more if volume is low) of the same amount, then the ARM would certainly be my choice, since the tools are better, the architecture is more flexible, etc. The fact that the architecture is newer, however, is no advantage.

In general, if the system cost and performance resulting from a given MCU choice is better, than the choice is easy.

Typical product lifespan, these days, is about three years. If the product works as intended, then maintainability isn't an issue if the product works properly. If it has to be changed within the three year product life, well, it's badly designed/debugged/tested and the people in charge, and the product itself, should be replaced.

RE



List of 71 messages in thread
TopicAuthorDate
2009 Embedded Systems Conference Trip Report            01/01/70 00:00      
   8051 dead?            01/01/70 00:00      
      I have heard that before            01/01/70 00:00      
         quoting Samuel Clemens            01/01/70 00:00      
   Not dead, but not sexy?            01/01/70 00:00      
   Ramtron?            01/01/70 00:00      
      Why is their part so much faster ...            01/01/70 00:00      
         it runs out of FLASH, not FRAM            01/01/70 00:00      
            Which rather kills their own claims for FRAM?            01/01/70 00:00      
            width of program memory            01/01/70 00:00      
               4 times as fast            01/01/70 00:00      
                  It might have to be wider            01/01/70 00:00      
                     Memory acceleration a growing problem            01/01/70 00:00      
                     Branch prediction?            01/01/70 00:00      
                        all these techniques...            01/01/70 00:00      
                           That's why I like the Maxim/Dallas parts            01/01/70 00:00      
                           In that case, BP was optional.            01/01/70 00:00      
                              yes, it's the good news, that it's possible at all            01/01/70 00:00      
   Dead? Keil (an ARM company) don't think so!            01/01/70 00:00      
      That's still their goal ...            01/01/70 00:00      
         native-mode environment            01/01/70 00:00      
            cross-compilation is king            01/01/70 00:00      
            I don't follow ...            01/01/70 00:00      
               Follow            01/01/70 00:00      
                  I still don't know what you meant, Andy ...            01/01/70 00:00      
                     Cheap native compiler normally equals cheap cross-compiler            01/01/70 00:00      
                        I'm told GCC is about the best there is ...            01/01/70 00:00      
                           It makes little sense            01/01/70 00:00      
                           I'm told that it may not be...            01/01/70 00:00      
                     Now I'm confused!            01/01/70 00:00      
                        I guess it's just a matter of preference.            01/01/70 00:00      
                           Still unsure what your "it" was            01/01/70 00:00      
                              the IT is in the full implementation.            01/01/70 00:00      
                                 $40 evb's for a fully-implemented ARM            01/01/70 00:00      
                                    Normally compiling on a PC            01/01/70 00:00      
                                    Will $49 do?            01/01/70 00:00      
                                       €39            01/01/70 00:00      
                                          STR9-comSTICK            01/01/70 00:00      
                                             Well, the goal is to leave the PC out of the picture            01/01/70 00:00      
                                                rephrase            01/01/70 00:00      
                                                   I must admit, that is quite true ...            01/01/70 00:00      
                                       Well, not quite ...            01/01/70 00:00      
                                          Full complement            01/01/70 00:00      
                                             Careful, now!            01/01/70 00:00      
                                    What I meant ...            01/01/70 00:00      
         ... which is in turn good news for the chipmakers...            01/01/70 00:00      
            I just bought a new hot-water-heater that's 'net free            01/01/70 00:00      
               yes, we know you love that old stuff ;-)            01/01/70 00:00      
                  old-stuff?            01/01/70 00:00      
                  W2            01/01/70 00:00      
         Windows for ARM...?            01/01/70 00:00      
            EEEEEEK!            01/01/70 00:00      
               I knew you'd like it!            01/01/70 00:00      
            There's already a Windows for ARM.            01/01/70 00:00      
   ARM for cheap            01/01/70 00:00      
      Cheap ARM vs Expensive 8052            01/01/70 00:00      
         The cheapest ARM chips can be had for $1            01/01/70 00:00      
            For hobbyists            01/01/70 00:00      
         It's getting to be that way, but ...            01/01/70 00:00      
      Yes, that one is pretty close            01/01/70 00:00      
         Don't lock down memory size - just think "good enough"            01/01/70 00:00      
            No, it should NOT be just that ...            01/01/70 00:00      
               All depending on product and actual product usage            01/01/70 00:00      
   TI buys Luminary            01/01/70 00:00      
      Detail (updated)            01/01/70 00:00      
      What's it like to be taken-over by TI?            01/01/70 00:00      
         Burr-Brown            01/01/70 00:00      
            Another update for your "family tree"...            01/01/70 00:00      
               TI had benchmarq too            01/01/70 00:00      
               who ate whom            01/01/70 00:00      
                  Benchmarq -> Unitrode -> TI            01/01/70 00:00      

Back to Subject List