For most operations that can be done in one machine instruction on a 32-bit CPU like Cortex-M7 (which also includes support for floating-point numbers), you need to write a complex sequence of instructions for 8-bit CPUs like Z80, 6502 and the like, where you do not have even a multiplication instruction, much less more complex operations.
To be able to write performant programs on ancient 8-bit CPUs requires much more knowledge and experience than on modern 32-bit CPUs, because you need to implement in software many algorithms for things that are done in hardware on the modern CPUs.