[speleo_engr]

Totally agree. I've been making the same argument elsewhere on this thread. People are disagreeing with saying things were simpler back then. I don't know how anyone can have fond memories of a 16-bit CPU with segments or bank switching compared to a modern flat model and think it was simpler.

[cesarb]

> I don't know how anyone can have fond memories of a 16-bit CPU with segments or bank switching compared to a modern flat model and think it was simpler.

"Fond memories" is easy to explain if that's what you grew up with. As for simpler, let me play devil's advocate for a bit: our "modern flat model" looks simple until you find out it's not really "flat". The 8086 model is basically "(segment << 4) + offset", while the "modern flat model" is actually a multi-level table lookup.

[planteen]

Yeah I guess that explains the fond memories part.

Some 32-bit parts are totally flat. Like low-end ARM parts with SRAM and no MMU. Compare that to a 8-bit or 16-bit PIC microcontroller where you need to bank switch to have a usable amount of memory for your application and its heaven.

But yeah, I see what you are saying. Still, pulling the wool over someone's eyes doesn't seem so bad to me about virtual->physical memory and TLBs as it does making them jump through a distinction between pointer types, but maybe I am in the minority on that.