Hi Andy,

An interesting topic you've raised here (popcorn anyone? ;-)

I suppose another way to ask your question would be: Can new variants like the ARM cortex M0-M3 architectures beat the 8051 core at it's own game i.e. that of a reasonably well-rounded Byte, Nybble and Boolean I/O processor!

On the comparisons between an ARM and 8051: Okay, let's try a instruction cycle-for-cycle comparison: I for one would be impressed if any current ARM MCU could do @ 2MHz what I was able to get an 89C51 to do @ 24MHz, that is, create a space-invaders style TV game identical to the one I burned into a 89C2051 over five years ago.. http://www.8052.com/forum/read/95151

The limitations on the ARM CPU are to be same as were imposed on the 89C2051, just to be perfectly fair:

1.) CPU has access to two system timers and and a 2MHz SPI channel (or UART channel running in shift register mode)
2.) CPU core system clock i.e. after the PLL MUST not greater than 2MHz!
3.) Code MUST FIT in 128 Bytes of RAM and 2048 Bytes of flash ROM!

Sure, I'm guessing even a luminary 100 series 'MCU' could do it given the resources available to it, but what if it were given the above constraints?

Also a side comment/rant about the ARM if I may: I would compare what they did with their Cortex-M0 to say, what Intel could have done (or hopefully would still do, wouldn't hold my breath though) with their i386 architecture, by only implementing a useful 386 opcode subset to produce a really low-power '386-compatible' core that runs in 'real flat mode' or whatever, complete with 32-bit only IN/OUT instructions and all and then call it a
'microcontroller'.. :P


Regards,
Valentin