This questiion goes to compare MC9S08 and ColdFire in speed.
How longer does it take a ColdFire core to execute a byte arithmetic instruction given that it works in 32 bits? I mean, in the 8-bit microcontroller, adding two 8-bit variabes is just that; but in a 32-bit microcontroller it needs to complete he 32 bits for each opperand firts, right? Should I give the ColdFire a higher bus speed to achieve the 8-bit variables addition in the same time?
Can anybody clarify this for me please?
Thanks,
Mariana
On Coldfire there's almost no difference to add 32 bits or to add 8 bits. Yes, sometimes overhead instructions must be added to sign extend 8 bits to 32 bits or chop the result back to 8 bits. But certainly any kind of arithmetic is faster on Coldfire.
The Coldfire has a 32-bit bus whereas the 8-bitters usually have 8 bit busses, and that means the 32-bitter can read four times as much data at the same clock rate. The 32-bitter may also be running at a higher clock speed than the smaller one (the one I use runs at 240MHz and can issue one instruction per clock) and also has a cache so it isn't slowed down by reading instructions and data from the FLASH memory, which might be limited to 100ns or so.
Tom