Erik,


MCU people count clock cycles and bytes. A fast clock is 100Mhz. Wire busses rule the roost. You probably have a ~1Ghz PC on your desk to do development with, a few hundred mb of RAM, half of which is used by Windows, and a single slow disk. When you type "make", your build is done before you can reach for your coffee cup, and that's just how you like it.


At work, my smaller systems have 2 - 4 3Ghz CPU's and 4Gb of memory, and a 12 drive array. The larger ones.... 8 core 32-hyperthread CPU's, 32+ Gb of memory with packetized 288-bit wide crossbar interconnects, and a 4Tb FC/AL smart RAID array with multiple controllers, each of which is more powerful than your desktop. We have 100+ CPU monsters, but I haven't found a use for them. My product has an automated daily build schedule, the compiler time alone is several hours on distributed machines.


Needless to say... Your version of "works well" and mine are somewhat different, and in general there's a substantial difference between your mindset and my mindset when thinking about performance. I'm here in part because I want to understand and keep in touch with your mindset, and apply the gems of wisdom to my space. That, and these little '51's are loads of fun. :-)


The reality is, most user's experience isn't tainted by either of these extremes.


So why doesn't your compiler pre-calculate 2*2 and store it as a constant? :-)


Rob