Why didn’t the bit war continue into the seventh generation? Why didn’t the amount of bits double to 256 bits like they did in past generations? Any insight into this would be appreciated.
In: 609
It became less about 32 and 64 bit, and more about which system had the exclusives you wanted. Or how they handled 3D games, whether they had online capabilities.
It’s like how cell phone plans used to be measured in minutes, then minutes were unlimited and they made money from texts, and now texts are unlimited and they charge for data. The target metric changes as new capabilities are introduced.
Primarily older game consoles had a single CPU for both graphics and game engine. So how many processing “bits” the single CPU had, was an easy way to measure it’s performance.
With modern machines, they have multiple chips that handle audio, graphics, game engine (CPU) – as well as memory amount and memory bandwidth – and all modern machines are 64 bit. So console manufactures had to find a new way to differentiate performance and started using TFLOPS (Teraflops) because it was a summary of many calculations the machine could do across all the different components.
Neither “bits” nor “tflops” is (or was) an *accurate* way to measure performance, but it was a *simple* way to make comparisons, and so that’s what the industry used.
**Source: Was one of the causes for the change in how performance was measured.**
People already commented on how more than 128 bits is probably useless for addressing. While that was the main driving force for increasing instruction width in the past, wider instructions (256, 512 bits) are still useful for vector (simd) instructions.
With the lowering cost of system memory, bits were no longer an adequate measuring stick for consoles since everyone could afford more than they could use at the time. So we started talking about other aspects of consoles. Like exclusive titles and whatnot.
There’s one more negative to using 128 bit over 64 bit (and 64 over 32) that I didn’t see mentioned.
It instantly doubles the size of all the references/pointers. In addition unless you take special measures it also doubles the size of all integers.
So just by increasing the size of the default memory unit, you instantly increase memory use of your game, often around 50% without any other changes.
That’s why in the years past you’d see recommendations to run the game in 32 bit modes if you had low memory in your PC.
Imagine you are in your house and you are trying to figure out where to put stuff and how to organize it. If you live in a studio apartment you really don’t have much room so you don’t have to explain much about where things are but you also don’t have much room for stuff. You get a promotion and move into a 3bd 3bth, now you need to explain your bed is on the 2nd floor in the third bedroom. You now have enough space for your stuff so you are fine.
Eventually you get married and have kids! Now you have more stuff so you move into a mansion with 10bd and 10bth so explaining where things are takes more time, 4th floor, 2nd bedroom or 3rd floor living room. This is kind of what the bits were like in a very loose sense. Having more bits is like having more bedrooms to put your stuff. The problem is that if you expand to a 30bd 30bth house it’s going to be pretty useless if you don’t have enough stuff to fill it. You could in theory own multiple cars but you only have 3 and the rest of the space is useless. In fact there is even a cost so to speak in having all that extra space as you still have to pay property taxes on the rooms even if you rnt using them.
For now 64 bit computing is pretty darn good for the near future that there just isn’t yet much reason to upgrade to a bigger house. Most software doesn’t hit the limitations of 64 bits so they don’t need to go through the moving pains of transferring their shit from one house to the next. This may change some day but for now what we have is plenty
The bit tells you how much RAM each processor can use at once. 16 bit = 64 kb, 32 bit = 4 GB, and 64 bit = 16 EB. That’s exabyte or one billion gigabytes. A ps5 has 16 GB or a billionth of the 64 bit limit and until you pass that 16 EB limit a 64 bit processor is effectively the same performance as a 128 bit processor, or a 256 bit one, or a 512 bit one.
The funny thing is there’s a decent chance we will
never need to meaningfully use 128 bit processors at all. Also the next step above that, a 256 bit processor, could potentially use enough memory to index just shy of every atom in the universe at once.
Basically it became impractical to squeeze more bits in, even just for marketing. So they gave it up and just used the fastest CPU and GPU chips they could get for the target budget even if it is less “bits” than the older ones.
Does anyone remember the System Wars in GameSpot forums? I was really active PS2/Xbox/GameCube Era 😅 Sony fanboys were called cows. Xbox fanboys were called lemmings, and Nintendo fanboys were called sheep. 🤣OH, the memories 🥹 mid 00’s were something else
Imagine that “bit width” in a processor is *a little bit* like the ability to attach more wheels to a vehicle.
Going from one to two wheels is great! Many people can ride bikes that didn’t or couldn’t ride a unicycle. Two to four is a revolution again – now we have cars! Four to eight gets us some heavy duty trucks. Sixteen wheels, now we have full semi trailers with redundancy.
32 wheels…um well, hard to say what we’d do with that. 64 wheels? You’d lose everything to rolling resistance.
So it is with 64-bits. It is enough to use all the RAM we have figured out how to use so far. Adding more “bits” by itself does not increase performance, just as nobody benefits from 32 wheels on a truck yet.
Latest Answers