Is bitness really that important how they say?

Is bitness really that important how they say?
Dreamcast was 128-bit, and no personal computer is 128-bit now.

no

Technically its 64-bit. Like how technically the n64 is mostly 32 bit. Adding extra processors doesn't magically make it 128 bit.

Do you even do the math, man

No it wasn't you fucking tard.

I don't think you get just how big 128 bits is. It's gonna be a long time before we even reach the limits of 64 bits

Fuck off retard

dreamcast GOAT

>Dreamcast was 128-bit
lolno

From what I know the Dreamcast had a 128-bit graphical processor but a 16 bit processor tagged on

Who even cares about the Dreamcast anyway?

AMD64 only uses 48 bits for memory addressing as is which allows up to 256TiB of memory to be addressed, we're a really long way away from even hitting that limit.

48bits is for the memory addressing, the CPU can still handle instructions that are 64bits wide just fine making all AMD64 processors 64bit.


Also Dreamcast used a 32bit processor it was not 128bit or even 64bit.

In the condition of consoles the "bitness" is usually marketing gimmick and not an accurate indication of true "bitness"

Strictly, the "bitness" of a CPU is determined by the maximum number of bits in the CPUs address space.

The N64 was actually 32-bit (only had a 64-bit accumulator for math which wasn't really useful)
Most 8-bit microcontrollers are actually 16 bit
The Dreamcast was 32-bit (only had a 128-bit floating point unit inside of the GPU)
The Intel 8080 (reported as 8-bit) actually has a 16-bit address space

To be accurate, these companies should honestly report the true "bitness" of their systems
> Number of bits in the address space
> Number of bits in accumulator (math co-processor, both floating and integer arithmetic)
> Number of bits in GPU address space
> Number of bits in GPU accumulator

Current AMD64 CPUs (Intel & AMD) do indeed only have 48 physical address lines. Programs written for them are however compiled with the full 64-bit address capabilities. Indeed, these companies should also be reporting these facts.

128-bit seems like a disservice.

>want to store a small counter like player's lives, capped at 99
>forced to use 16 fucking bytes
>have to use bit packing to force multiple values into the same 128-bit word for efficient memory usage

Nothing wrong with bit-packing multiple values in to registers and such.
Of course, we are talking modern devs. They don't know what registers are.
All they know is jQuery, wizards and templates. Registers? Register for newsletters to the new hottest library updates?! SURE!

It's not like it is some mega hard process. Concatting and deconcatting is pretty trivial in most languages.
Even JavaScript. Although I am not 100% sure if wasm lets you work with registers. I've not even cared to look at the spec. I'll care in 10 years when it is beta. Maybe.

Doesn't that just make things more of a pain in the ass when it comes to modifying values in that address space though?

Like every time you want to update something you need to update the whole thing. And if two values change near the same time you might get an overwrite as it tries to update.

CPU uses some address translation table I believe.

It'd help if you actually knew what "bitness" is.

do you? they called it that cause it ran float point

That's not how it works. I can tell you're a 12 year that thinks he's figured out technology. You'll look back on this in a few years and cringe.

Dreamcast was kill because it would have made bitmining trivial.

Now you know the truth.

they said this about 32 bits

just wait for exponential complexity

x86-16 was introduced in 1978.
x86-32 was introduced in 1985, seven years later.
x86-64 was introduced in 2003, eighteen years later.

We're currently 14 years into x86-64 and the limit of it is still so far off that we don't even max out 48 bits of addressible memory yet.

Don't worry about random bit counts in computer architectures that you don't understand.

That's not how it works.You could store the counter as a single byte in memory. When its fetched into a register, it will be extended to the register width, which doesn't have to be 128 bits. x64 has both 32 and 64 bit registers.

Depending on how you measure it, its 8,9, 16,32 or 128 bits.

"Bitness" was a decent classification of the generation of a computer while it was growing. Since its static now, it obviously isn't any longer.

Yeah, people that don't seem to understand how large the 64-bit memory address space is

plus there number of extra instructions on a modern CPU (SSE, AVX, etc), plus the rise of GPU compute have essentially guaranteed 64-bit will be with us for a very long time.

Never was.
The 8bit PC Engine was faster than the 16bit sega genesis at sprite positioning (the very thing that slowed down consoles back then), so from the start it was fucked.

although encryption keys are 256 and even 2048 bits
but maybe I misunderstand

Meh, in that sense, modern CPUs are already 256-bit, taking the new YMM registers and instructions into account.

Why are they not measured in bytes instead?

8 bytes rather than 64 bits

Marketing.

Saturn was marketed as 64bit at the japan launch, since it had two 32bit processors. Additionally it has a DSP that can compute 48bit numbers.

Most Dreamcast games, if not all, used 16-bit framebuffer. Possibly because it had to have every draw command for a frame in memory before it can draw the frame, so they'd try to save space with the texture and framebuffer colour depth to get larger poly count - but this is just a guess on my part.

N64 did have 64bit cpu instructions, but they were slower, so no point in using them at all.

The Motorola 68k in most 16-bit computers and consoles had a 16-bit ALU and external bus, but internally it was 32bit otherwise.

The SNES was a 16bit console but its CPU was 8-bit except for the ALU and the internal registers.

The Dreamcast had a 32-bit RISC CPU.

What the fuck does a game console do with 48-bit sized samples in a DSP, image processing?

no, the personal computer is 512+ bits
look up SIMD, retard

The DSP was meant to be used for T&L, and for that, higher precision is good.

That's pretty neat, too bad the saturn sucked at getting 3d games

With nanotechnology, we could make computers that have mole quantities of transistors and memory elements. Maybe then we'll need 128 bit processors

eh. Intel has 512-bit instructions on current CPUs, so you could say it's 512 bit, in much the same way as the Dreamcast was 128 bit.