Andy Warhol, Big Mac, iPhone: It was a grand American epoch. But it’s over. Europe must finally emancipate itself – just not as awkwardly as Jürgen Habermas might like.
I’m not a developer, but if I’m not mistaken, every compute still uses the Von Neumann architecture. Someone please correct me if I’m wrong.
The vast majority of modern computers use the same hardware mechanism to encode and store both data and program instructions, but have caches between the CPU and memory, and, for the caches closest to the CPU, have separate caches for instructions and data, so that most instruction and data fetches use separate buses (split-cache architecture).
Most today are “modified Harvard architecture”. Instruction fetch and data fetch are independent of each other (Harvard architecture) but they don’t have separate memories, just one (the modification).
Honestly, things have moved on so far now that modern CPUs only resemble Princeton (Von Neumann) and Harvard architectures in very loose ways. Out of order execution in particular rewrote the rule book.
Not really because things are such a mish-mash of good ideas by different people. Over the past few years I’ve been collecting and reading a lot of the seminal papers and I’ve probably got 20-25 of them where ideas were introduced. A lot more are internal developments held secret inside companies like Intel or AMD.
I guess if you tried it might be something like “Super-scaler out-of-order speculative pipelined modified Harvard architecture” but that’s just scratching the surface. The original concepts are still there, but there’s a lot of work on top.
Edit to break that down:
Pipelined: instructions can be issued before the result of previous instruction is known, if there is no dependency. Allows one multi-cycle instruction to be issued per clock. This, in turn, allows the cycle time to get shorter.
Speculative: instructions can be issued before it is known if they will actually be executed. Incorrect Instructions are killed before they complete. Avoids conditional branches causing big stalls.
Super-scaler: It has multiple execution pipelines so more than one instruction can be issued per clock.
Out of Order: instructions don’t have to execute in program order if the are not dependent on each other. Allows the processor to work on future things whilst waiting for a result to become available.
I’m not a developer, but if I’m not mistaken, every compute still uses the Von Neumann architecture. Someone please correct me if I’m wrong.
Most today are “modified Harvard architecture”. Instruction fetch and data fetch are independent of each other (Harvard architecture) but they don’t have separate memories, just one (the modification).
Honestly, things have moved on so far now that modern CPUs only resemble Princeton (Von Neumann) and Harvard architectures in very loose ways. Out of order execution in particular rewrote the rule book.
Is there a name for the modern architectures?
Not really because things are such a mish-mash of good ideas by different people. Over the past few years I’ve been collecting and reading a lot of the seminal papers and I’ve probably got 20-25 of them where ideas were introduced. A lot more are internal developments held secret inside companies like Intel or AMD.
I guess if you tried it might be something like “Super-scaler out-of-order speculative pipelined modified Harvard architecture” but that’s just scratching the surface. The original concepts are still there, but there’s a lot of work on top.
Edit to break that down:
Thank you for the breakdown. Very interesting.