So there no "x86 only" software.
Mmmm, depends, this it's a pretty broad assumption, like to said Java runs everywhere w/o recompilation, likewise you may use either CPU specific instructions which define an algorithm behavior, encryption and big endia/Little endian it's just one on many examples, we don't talk about web services transactions, but those doing math or Bitcoin will illustrate you a bunch of utilities (like tesselation) that requires platform custom code and can't translate "on the fly'.
usually write "High level programming language" like C/C++/ObjectiveC/Swift/Java/Kotlin.
Likely with Java/kotlin exception all those provides "optimization flags" so the llvm compiles analyze the code and by itself implementation CPU extension, a common situation are parallelism and maths, I'll need to write a think book with the few of them.
As others have noted, this will be a staged roll-out, with the most powerful Mac models (the ones being used in corporations) being the last to migrate off of native x86 CPUs.
I don't buy Apple switching to ARM as they did when adopted x86 from (actually conceptually superior) powerpc, but maybe cook has a plan "just in case' the iPad loses its magic, like M$ rollout in parallel an ARM optimized macOS-arm platform only running catalyst applications as alternative safeguard strategy to an hypothetically declining iPad market, likewise the Rich iPad App store ecosystem brings macOS-x86 fresh air of new apps reverting macOS App store sad decline.
An ARM Mac won't be an alternative to an Mac pro iMac 27, MBP even high end Mac Mini, but would be ideal for low-end market (or iPad switching users, as also a side effect from catalyst is users abandoning iPads to enjoy more comfortable work-oriented device).
CISC typically uses microcode to break down these complicated instructions into simpler RISC-like instructions
Not always sir (I follow your post, but I won't assume an stake on your claims), I assume you likely retired, otherwise you should aware what Intel and AMD do adding more and more Gates (an modern x86 core easy cuadruplicate am ARM core).
Code execution optimization takes many strategies, like decode cisc to risc and then implement an optimized execution pipeline, is something working since i486 (and inspired in part by WISC CPU -itanium-), but modern CPU go beyond binary decomposition and recomposition, while predictive instruction sequence can improve IPC in pipeline execution models, newer CPU also execute some complex instructions deliberately in an asic often in an single cycled, even ARM CPUs follows this strategy but while cisc/WISC CPU can implement it deliberately ARM only does in pipelines as part of it's predictive execution branch, it gives cisc/WISC CPU an little edge over ARM, an edge that requires binary optimization sometimes.
Considering we are at a blink to leave behind the Von Neumann architecture (mostly thanks to ml/ai/cues), reinvent the scalar CPU is not a long term paying strategy.
I have few questions, for you sir, with due respect, where are macOS leaks pointing out apple is testing ARM CPUs beyond the PCH/SMC (aka t1/t2) controller? Why we have some AMD Zen APUs there?
You are familiar (and likely meet they) with Jim Keller and Mike Filippo? ARM CPUs? Right, AMD Zen... Right , yes Keller was Filipino chief while crafting Zen at AMD, if Apple can hire someone who knows how to improve or develop AMD - Zen licensed architecture, is Filippo, he knows Zen better than ARMs, despite his work crafting an x86-feature/performance competing ARM CPU, it was just monts of work, while Zen means years of experience.
Licensing Zen from AMD had Sense, as the CPU platform implies almost negible modifications to macOS (some of which seems done, ask Hackintosh-ers now capable to run macOS on AMD , as the never could until 2 yr ago).
I don't buy the ARM platform would bring macs that big advances you claim, every Von Neumann system implementation has its days numbered, do you want Innovation? Go hybrid, asic/fpga/nvram/many micro cores/integrated storage, in the future system ram, CPU, GPU, TPU, flash storage will reside in the same chip complex, you need more power, you add more of these hybrid complex, not just more ram etc.
My regards sir just in case you're who you claimed to be.
es.wikipedia.org