@Talloran @tuxcrafting except it wasn't exactly x86 that become popular, it was some proprietary software build for x86, at which point as a CPU manufacturer, even if going out of your way to provide an optimizing compiler for your shiny new architecture, you were on the whim of the software copyright holder to (allow you to) rebuild the software, because software was what sold hardware(still is). So, ironically, for hardware manufacturer to remain independent from the software monopolists, hardware designers had to stay backward compatible with that software, shamelessly pushing the rhetoric of it being some sort of a fundamental problem that they are so good at solving, even when GCC proved that we can do much better. That's how deeply proprietary software culture messed everything up, and it is a complete failure by any metric, other than extortion.
@tuxcrafting
they don't go hand in hand, proprietary software is kicking ISA compatibility forward under a gunpoint, and every time it stumbles it gets a whipping to get it to go faster, while crying and moaning, barely alive from the numerous wounds it sustained.
"it had to improve while keeping backward compatibility, so the instruction set couldn't really be changed" only true for proprietary software, and this is why first things you see ported to new architectures are usually foss (other than maybe manufacturer's/designer's own stuff that is prepared before release).
Even just assemblers can do a lot to make life easier for you, so source vs binary distinction existed before higher level languages became popular. Disregarding that, they did become popular long ago, so whatever assembly code remains doesn't really explain why x86 is still "today's most successful line of CPU chips"
