I didn't bring up the bit-banging. You're right, of course, there are limitations as with any hardware. The fastest rate at which you'd be able to update a bit-banged port on an external-memory-interfaced 8255 attached to your 'F12x-series MCU would be a little above 4 MHz.

However, you've once again attempted to evade the original question, which was regarding how much the rare occasional access to a 230 ns external bus would slow down your code. I did some arithmetic and showed how insignificant it was, but you've chosen to ignore that.

What you've got to consider, though, is that not everyone has the option of building a new PCB for every little job. There are many tasks that are attacked every day by using a COTS product, and, if you engage in that sort of work, as many people, including me, do, then you have to settle on a product.

While I'm sure that you clear your boss' shelves of everything that's over two weeks old all the time, some people don't do that. If you walk into many facilities, you'll still see equipment from DEC. They went away over a DECade ago, yet their products still work just as well as they ever did, though I never learned to like 'em. Many industries, notable aerospace, like "tried and true" much better than "bleeding edge" for their products. The power industry is even more conservative, as they simply hate change. As a result, many power facilities, regardless of how they're fueled, have 1980's generation hardware in house, and spare parts with date codes from back then in stock to facilitate maintenance.

I'd venture to say that there are more 8255's out there still doing the jobs for which they were originally intended, than there are SiLabs MCU's, including those that are still just inventory. They still do what they're supposed to do. You can't deny that they work. It's not their fault that you don't know how to make them work. If you find them distasteful, then don't use 'em. It's your loss.

However, you've no business telling someone else what to do or not to do based merely on your own ignorance and lack of imagination, particularly as you seem to fail to recognize even how to use the parts that you claim to use every day.

Go back. Read the datasheets. Ask yourself, "Why would they include such a feature if it makes no sense to use it?"

I'm puzzled that you seem to believe that a speed reduction that's many orders of magnitude below what's produced just by the variation between commercially available crystals (at best, 20 ppm) is going to interfere with a reasonbly designed project. Perhaps you can explain your logic.

My position remains that you should not tell people NOT to do something just becuase YOU don't know how to do it. Further I don't believe you should tell people that they have to build a circuit board just because they otherwise can't use YOUR MCU of choice. There are students and hobbyists throughout the world who'll never need to build a PCB. Why should they use SMT devices just to please you?

RE