So far, I’ve come across these three terms:
- Princeton (John von Neumann) architecture
- Harvard architecture
- Modified Harvard architecture
Is that it? What about some of the other alternatives?
You must log in or register to comment.
Depends how far you stretch the definition. Those listed are a subset of RASP machine models, which are a subset of turing machines, which is a subset of finite state machines, but as far as “general purpose computing architectures” that’s about it as far as I’m aware… Unless you get into weird application specific stuff like differentiable neural computers.
You could also go the other way and look at how the basic principles of those architectures get expanded on IE Von Neumann -> ARM -> Advanced Microcontroller Bus Architecture (AMBA) -> Advanced eXtensible Interface (AXI) -> etc.
Would a quantum computer count as a finite state machine? If not, wouldn’t it be an example of something different?
Tbh a little bit outside my wheelhouse but Lance Fortnow had a proof showing that all the implementations of quantum computing could be modelled using a Quantum Turing machine and are more complex forms of probabalistic automata. It’s technically not, because a finite state machine can also be called a deterministic finite automata where a probabalistic automata is a Nondeterministic finite automata.
So no… But also mostly the same principle just with fuzzy logic.
Neuromorphics are a big one that will soon have a large presence. This also sort of encompasses tensor units and other AI specific hardwares. Maybe take a look at ASICs and unconventional computing.
That’s definitely not it but it’s very difficult to find research (especially ongoing research is very rare) or even information on alternatives to the currently dominant (due to monopoly) von Neumann, binary, register machine, instruction-set-having, etc architectures.
You might try looking into dataflow, logic-based (like computers that do Prolog), or transport-triggered architectures. Or Lisp machines too.
