Most of what I did back in 80's and early 90's was writing software for the hardware, I designed and built.
I worked on the 8031/8751, 8085, and the Z80. On one occasion, I faced a challenge to build a Private Automatic Branch Exchange (PABX). The prototype was 8-Subscribers by 2-External lines. I planned to leave the majority of the tasks to be executed by the software to minimize hardware. At that time I was not well acquainted with Object Oriented programming, and everything had to be written in assembly. I realized from the start, in order to reach my goal successfully, I had to separate procedures/functions from data. Each subscriber/Extension had its set of data and flags as a separate entity. I needed a processor with a fair amount of ROM and RAM space, in addition to flexible addressing modes, comfortable room of internal registers, and unlimited stack space. There was critical timing involved, since everything was running in real time. At that time and for that specific project, the logical choice was the Z80 CPU.
All I had at the time was a PC-XT Amstrad running DOS, a Z80 cross assembler, a terminal app and a general purpose editor with a multi window feature. The target hardware was running a simple boot-loader/monitor program I wrote my self ... and that's it.
In my opinion, it is not the Z80, 6502, ATMega, or PIC. True, you have to make a choice on what resources you need for any specific task. However, what really matters here is the ecosystem behind the hardware of choice and whether there is an active community supporting the cause. I am very astonished here to find people still interested in the Z80!
If this enthusiasm builds up further then I won’t be surprised to find companies starting to re-invest in the Z80 and its chip-sets.
Moreover, Instead of inserting inline assembly code on a compiler, one day the trend might be; inserting inline high level code of the language of your choice in an assembly program.
