Dear homebrew developers why the original Xbox was able to emulate N64 games flawlessly and a twice faster CPU on a Raspberry Pi 3B/3B+/4 (and far more RAM) still can’t?
In: Technology
Mainly because the CPU clockspeed isn’t the only factor to consider.
E.g A bus and a motorbike can both travel at 100kmh but at the end of the journey the bus will have transported 50 times more people and their luggage.
It’s the same with CPU’s. Different designs (essentially known as architecures) can do more than others. They all perform various calculations on data (essentially known as instructions) but some can have more complex instructions to choose from and so can do more ‘work’ in less time.
That’s basically the difference between the Xbox CPU and the Raspberry Pi. It’s the CISC (complex instruction set computer) architecture versus the RISC (reduced instruction set computer) architecture. The Xbox uses a Pentium III (CISC) and the Raspberry Pi 3b uses a Cortex A53 (RISC).
If you want to get a bit more technical we can take the example of multiplying two numbers. One number is stored in memory location A, the other is stored in memory location B, we want to multiply A by B and store the result back in memory location A again.
With the CISC architecture of the Pentium III we just have to tell the CPU to do something like :
multiply a, b
That’s one complex instruction that takes one clock cycle to complete. The Xbox CPU runs at 733MHz so it can do 733 million of those multiplications in one second. (That’s not exactly correct but it’s ok to think of it that way for the sake of the example).
With the RISC architecture of the Cortex A53 we have to tell the CPU to do something like :
load a
load b
product a, b
store a
That’s 4 simple instructions that take 4 clock cycles to complete. If the Raspberry Pi 3b CPU runs at roughly double the clockspeed of the Xbox CPU it would still take twice as long do that multiplication task. (Again that’s not exactly correct but think of it that way for the sake of the example).
It gets even more technical when you take other aspects of the different CPU architectures into consideration. You can get into things like how many instructions can each architecture execute at the same time (known as parallelism). That’s not as simple as how many cores are available, it’s to do with how each core schedules and arranges the instructions it’s working on at any given moment. It does that because instructions and memory accesses take time to complete, and the CPU can be doing other work while it’s waiting for those things to complete. The Pentium III has a superior design in that department compared to the Cortex A53 which further increases the amount of work it can do in each clock cycle.
Or you can talk about the size and performance of the cache memory (very fast memory inside the CPU that stores the most recently used data and instructions). That is used to avoid having to load and store the same data from the relatively slow main memory (the RAM). That also allows work to be done more quickly and the Pentium III is better than the Cortex A53 in that regard too.
So … as you can imagine all that extra complexity in a Pentium III makes it more capable, and it also makes it larger and more expensive to manufacture. But that’s basically why an Xbox CPU is better than a Raspberry Pi CPU even though it’s clockspeed is half as fast, and also why a Raspberry Pi CPU is so much cheaper.
Latest Answers