Lack of backwards compatibility every time Apple changed their OS or processor pretty much ruined my life back when I was trying to write games on the Mac, because they coincided with downturns like the dot bomb and housing bubble popping (in that case just after iOS arrived). I was so beat down trying to survive at life that rewriting everything I had just written the last year for the new hotness became too much of a burden. I worked a bunch of dead end jobs instead and wasted whatever potential I might have had. Now midlife has hit and I've generally let all that go, but it still bothers me thinking about what might have been.
That said, Intel has sat on their hands for 15 (I would argue 20) years, and so it's unsurprising that Apple is ditching them. I remember seeing 3 GHz processors sometime around 2003-2004. Very little has changed since then. We have faster memory busses now but we've generally lost almost two decades of Moore's law. Before that, processors got twice as fast every 1.5 years, so 100 times faster every decade, which would be performance equivalent to a 30 THz processor today.
Note that where progress HAS happened is video cards (GPUs). So I'm somewhat optimistic that if Apple disrupts the CPU industry, we might see true general-purpose computation speed up rather quickly and break the 4 core, 1 memory bus barrier. I think low hanging fruit here would be 16 to 256 cores arranged in a grid, with the square root of that number of memory busses on an edge. With today's tech, we could have 1024 DEC Alpha cores with 32 memory busses for not much more than we're paying for an Intel i9 with 2 billion transistors (the Alpha had 2 million). Yes, I know it's not an exact comparison, but I have a computer engineering degree so this isn't the time to be pedantic.
General purpose computing could also disrupt the GPU and AI industries, because we could jump ship from the ever-narrowing niches of rasterization and neural nets, and move on to broader experiments in things like ray tracing and genetic algorithms. I had originally wanted to do that with FPGAs, but I've been burned out so long trying to keep up with the shortening attention span of tech that I had to let it go.
Hard to say if any of this will happen, but I just wanted to shed light on the kind of innovations we've missed out on under two lost decades in tech. This is the tip of the iceberg. An explanation for this is that customers want cheap eye candy, and prices have certainly fallen on track with Moore's law. But I've vote to finally see better performance again. Also I'd like to see the emergent effects of better processors, such as more use of parallelized higher-order functions and data driven/functional/declarative programming using something like the Actor model, piping data around with simple tools that do one thing well, borrowing techniques from UNIX and Clojure/Erlang/Go/MATLAB/etc.
All of that stuff can be good, but has tradeoffs. Longer pipelines result in worse branching performance, caching interferes with write-heavy code that's mainly about moving data (like for games), and so on. I feel that putting extra transistors towards large numbers of cores with short 4 stage pipelines (like in early PowerPC) would have been better.
This is one of the more concise benchmark comparisons, in this case having a 3.6 GHz i9 and 1.4 GHz Pentium 3 (released starting in 1999):
So this is 8 cores vs 1, at 2.57 times the clock speed. So per-core performance has increased:
(18892/299) * (1/8) * (1.4/3.6) = 3.07
A 3x fold increase in 20 years is admirable but 1/3000 what would have been predicted if performance had followed Moore's Law. To me, this indicates that per-core performance stopped really increasing sometime around 2005 at the latest. That's why fabs moved towards lower-cost mobile and embedded chips.
That said, Intel has sat on their hands for 15 (I would argue 20) years, and so it's unsurprising that Apple is ditching them. I remember seeing 3 GHz processors sometime around 2003-2004. Very little has changed since then. We have faster memory busses now but we've generally lost almost two decades of Moore's law. Before that, processors got twice as fast every 1.5 years, so 100 times faster every decade, which would be performance equivalent to a 30 THz processor today.
Note that where progress HAS happened is video cards (GPUs). So I'm somewhat optimistic that if Apple disrupts the CPU industry, we might see true general-purpose computation speed up rather quickly and break the 4 core, 1 memory bus barrier. I think low hanging fruit here would be 16 to 256 cores arranged in a grid, with the square root of that number of memory busses on an edge. With today's tech, we could have 1024 DEC Alpha cores with 32 memory busses for not much more than we're paying for an Intel i9 with 2 billion transistors (the Alpha had 2 million). Yes, I know it's not an exact comparison, but I have a computer engineering degree so this isn't the time to be pedantic.
General purpose computing could also disrupt the GPU and AI industries, because we could jump ship from the ever-narrowing niches of rasterization and neural nets, and move on to broader experiments in things like ray tracing and genetic algorithms. I had originally wanted to do that with FPGAs, but I've been burned out so long trying to keep up with the shortening attention span of tech that I had to let it go.
Hard to say if any of this will happen, but I just wanted to shed light on the kind of innovations we've missed out on under two lost decades in tech. This is the tip of the iceberg. An explanation for this is that customers want cheap eye candy, and prices have certainly fallen on track with Moore's law. But I've vote to finally see better performance again. Also I'd like to see the emergent effects of better processors, such as more use of parallelized higher-order functions and data driven/functional/declarative programming using something like the Actor model, piping data around with simple tools that do one thing well, borrowing techniques from UNIX and Clojure/Erlang/Go/MATLAB/etc.