256-bit systems WHEN!??!

>>56172480
give me a week

y tho

>>56172554
y not tho

We do not know name for such big number or how even address this space

that's a lot of processing

Y-you sure you can handle it?

>>56172657
I need to be able to render videos in an instant.

>>56172480
Because we're still nowhere our 2^64 memory address space, and just adding more bits isn't going to help anything.

Fuck naw, let's just go 4096

File: 1471609856515.jpg (48KB, 435x571px) Image search: [Google]

48KB, 435x571px

>>56172704
8192 bits or go home.

>>56172695
Look at this fag.
He doesn't even have a Microsoft Hololens 5D yet.

Get with the times gramps.

>>56172480
You sure as hell don't need that for memory addressing. So, y tho? 128 bit arithmetic instructions are already present on our 64 bit CPUs, if the need emerged, they'll add 256 bit ones as well.

>mfw i'll never own a 4Ghz+ 256bit bus and 128 core CPU with 128MB L1 L2 64mb and L3 32mb cache

>mfw It will never exist in your life time
;_;

I'm fine with my 86 bits system

File: 185290503.png (3MB, 10000x8640px) Image search: [Google]

3MB, 10000x8640px

>>56172824
>86 bits

>>56172480
You don't know what that means, do you?

>lolol who needs that much memory

AVX

>>56173285
If we apply Moore's law to memory (doubling every 18 months) we get 30 doublings from 1970 to 2015. Will somebody explain to me why are we so certain that in 50-60 years there won't be a comparable amount of doublings and that we'll get the same issues we had with 32 bit systems? Or is it just a "fuck that's a long time, somebody will figure something out" like the fucking Unix time?

>>56172824
ebin
b
i
n

>>56173455
Moore's law is failing though

>>56173523
intel or nvidia dont have competitors so thats why theres no moores law, just increse performance by 5-10% and add $100 to the price of the old one.

>>56173523
There's no physical reason for it to fail though. Current tech may be close to end and new approaches are required but they aren't within tens of orders of magnitude of what's physically possible.

>>56173583
Normal transistors can't easily get any smaller

>>56173583
id be happy if intel did take like 4 i7-6950X's and crammed in one chip. dont care if the die size increases. let it be 65nm again whatever.

what's the difference though its just more numbers

>>56174098
more raw power.

>>56174130
how do more numbers bring more power

>>56174146
>more numbers =/= more power
>1 core vs 400 cores

??

>>56174168
i dont evne know anymore

>>56174202
>tfw you will never see a processor clocked at 1THz...

>>56174237
http://www.extremetech.com/extreme/193343-darpa-creates-first-1thz-computer-chip-earns-guinness-world-record
it's only a solid state amplifier, but 1THz nonetheless

>>56174362
do one for home computers.

Wtf do you need a processor word size like that for? There is no perceptible need or advantage to doing so. We don't need to address memory space that large, there aren't come data types that could leverage that. Nobody needs it yet. One anon is right though, maybe in 50 years.

File: 1451229713151.jpg (261KB, 900x1350px) Image search: [Google]

261KB, 900x1350px

>>56172803

1MB cache L1 when?

lmao even nintendo had 64 bit console in fucking 1990 and computers are still stuck with 64 bit, fucking pc gamers btfo!

imagine the size of those register files

>the year 2072
>Call of Duty: Prehistoric Warfare released
>Still uses ID tech 3
>Game size is 3TB because of textures

File: Qouprsg.jpg (60KB, 612x441px) Image search: [Google]

60KB, 612x441px

>>56174480
>Game size is 3TB because of textures
>mfw still on my 1/1 connection

>>56173621
>current tech may be close to ends and new approaches might be needed

>>56172480
64 bit address space is enough

>>56174529
>there's no physical reason for it to fail
>""new approaches"" might be required, e.g. magic

A toaster is always enough.

Always.

We reached space with one.

>>56172729
>if the need emerged, they'll add 256 bit ones as well.
They already did

https://en.wikipedia.org/wiki/Advanced_Vector_Extensions

>>56174553
Space is shit, just some stupid rocks

>>56174516
>tfw game system requirements havent changed much
>To play in FULL FULL FULL QHD HD UHD 16K
>you need
>CPU: i5-4x9000 quad core cpu
>Ram: 32gb of DDR6
>GPU: Nvidia TITAN QWERTY
>Storage: Atleast 10TB storage

>>56173583
>There's no physical reason for it to fail though.
Moore's law is about economics, not physics. CPUs got faster exponentially because the demand for computing was growing exponentially.

Moore's law is failing becaus the market is close to being saturated.

>>56174557
Earth is just some stupid rock too

>>56172592
Performance comes to mind

>>56174557
>Space is shit, just some stupid rocks
So are you, just a stupid cock.

>>56174432
Literally never, dumbass.

>>56174553
We reached space with a toaster and created universes on more modern pcs

>>56174579
yet you shitpost on a $1000 PC

>>56174589
More like $10,000

>>56174589
true.
but after I'm done shitposting I can go do many more things.

File: 1467632046484.jpg (18KB, 708x627px) Image search: [Google]

18KB, 708x627px

>>56174565
>mfw raid 2 sssd isn't enough
>tfw have to buy some atomic cards to get above 200 fps

life is hell

That cost me like $2000 Space Dollars

>>56174607
but will you?

>>56174568
>>56174576
Triggered NASA employees detected
Enjoy knowing I'll campaign for you to be defunded
Maybe we can help OP and redirect the money to intel or something

>>56174599

int i;
cout<<"What's your PC's value?\n";
cin>>i;
i*=10;
cout<<"Fuck off, it's actually"<<i;

>>56174630
My battlestation is literally $10,000 though

>>56174623
>NASA
I read that as NSA.

DId you?

ITT:
People who don't understand memory addressing
People who don't understand vector width
People who don't understand integer length

>>56174566
Performance increased on an S curve, not exponentially.
We stopped putting transistors into logic because we have no idea how to spend them. We increase transistor count by adding cache, tertiary accelerators, more cores, and IGP.

The "failing" people speak of in reference to Moore's continually shifting estimation is due to cost per transistor increasing for high performance devices on bleeding edge nodes. Per core performance uplifts have tapered off due to diminishing returns. Making a core wider doesn't increase performance unless you can feed it, and its damn near impossible for us to fully utilize a general 4 wide int core 100%. 3 wide designs have the 3rd ALU so underutilized that a second logical thread can run on them.
SMT literally exists because this problem was envisioned decades before it came to materialize.

>>56172824
Quality kek

>>56174644
>Performance increased on an S curve, not exponentially.
Market growth is also an S curve

>>56174643
wouldn't be surprised if they were investing in machines to look at "other" space rocks

>>56174674
>wouldn't be surprised if they were investing in machines to look at "other" space rocks

>implying the NSA isn't just a ponzi scheme to get more moneys in the pockets of the director and the other beaureucrats

Literally nothing has happened with the NSA, it's a fucking scam.

File: 1343390600839.jpg (527KB, 543x810px) Image search: [Google]

527KB, 543x810px

>>56174674
>intelligent life is found
>the NSA places it on a terrorist watch list

At least we know we'll have one thing in common with alien life: No matter where you come from in life the government always tries to fuck you.

>>56173894
Xeon?

>>56172499
I'll get working on the logo

>>56172718
65535 or bust

>>56174819
>not even a power of two
the fuck kind of abomination would that even be

>>56174859
yeah it is (unless I'm wrong, can't remember my powers of two off the top of my head)

>>56174859
Its a power of 3.

>>56174859
2^16 - 1

>>56174876
It's an odd number you dork.

>>56174919
This is a safe space, don't call someone a name you wouldn't call yourself.

>>56174919
oh oops i think it's 65536, must have started at 0 or something
i can't function this morning

>>56174819
Seems a little confused. It would be the maximum address in a 16 bit system.
The 256 in the context of this thread would mean the same as 16 in this post.
https://msdn.microsoft.com/en-us/library/system.uint16.maxvalue(v=vs.110).aspx

>>56172824
can't wait for that sweet x86_128 system tbqh breh :^)

RISC-V supports 128-bit addressing because of a few companies that are starting to hit around 54~56 bits of addressable memory in HUGE databases, that might maybe need more than 64 bits in a decade or two. Going beyond that, however, seems a little counter intuitive. The vast majority of programs hardly even break out of the 32 bit address space, to the point where the only reason to use a 64 bit version of a program is the improvements made to x86 and ARM in the transition to 64-bit (both had a major overhaul of the ISA itself).

256-bit is overkill, and we might hit some physical walls in terms of how much memory we can reliably put into a RAM chip, and how fast we can access memory. Why does speed matter, you might ask? Well, as one of my professors had pointed out to me in a discussion I led on the benefits of RISC-V, the problem with having even a 128-bit address space is, "how much time would it take to actually access every physical address once?" Even the limits of 64 bits is pushing things.

>>56173894
>4 i7-6950X's and crammed in one chip
>quad house fire

>>56172480
what makes you think that the average user needs anything over 32-bit? if it weren't for the fascination with pics and videos, we wouldn't even need 16-bit.

>>56175597
>256-bit is overkill, and we might hit some physical walls
wait until we start using planetary databases

>>56175912
Asimov was right

>>56175912
>>56175930
>Asimov
I think you mean Orwell

>>56172695
You're an idiot

It's time to start using giant processors since transistors are barely getting smaller anyway.
We'll have to abandon shitty slow electricity and use light though.

>>56176008
Ditto.

>>56172824
lmao

>not 32768 bits yet
what are you, poor?

>>56174572
No it doesn't.

>>56174950
I'd call myself a dork, dork.

2^64 bytes ought to be enough for anybody

>>56174963
>oh oops i think it's 65536, must have started at 0 or something
wait did you actually count all the way from 0 to 65535?

>>56175597
>because of a few companies that are starting to hit around 54~56 bits of addressable memory in HUGE databases

that's 72 PB

what the shit

>>56175597
>256-bit is overkill, and we might hit some physical walls in terms of how much memory we can reliably put into a RAM chip
At 256 bits you hit thermodynamic limits of the universe

We will not be exhausting a 256 bit address space until our computers are made of something other than matter and occupy something other than space

>Cmd+F "PlayStation 2"
>0 results
I'm impressed, /g/. I thought at least one of you retards would claim that 64-bit machine is a 128-bit one.

>>56175597
it's still a fucking massive jump from 56 to 64 bit
like, 63 is only half way there

>>56176645
more like /g/ not caring about >>>/v/ crap

>>56176645
>ctrl+f
>"playstation 2"
>1 result
i knew one of you retards would mention it

>>56176653
2^64 is 18446 PB, which is about 1000 as much as the internet archive currently stores across all its machines

(Never mind the fact that they are not mapped into the same address space)

>>56176656
>implying 80% of /g/ isn't a /v/ leak
Nice joke

>>56176674
It was inevitable.

>>56176656
It's only /v/ crap if you're talking about the games on it.
The PS2 architecture itself is interesting.
Stop being blinded by your own retardation.

>>56176645
>cmd
>stupid comment
>>>/fa/ggots
>>>/lgbt/

>xbit-systems
>fixed size registers

It's not the 90s anymore granpa

>>56176705
>lolol macs r gay xDDDD
This is an 18+ website.

>>56176718
the architecture bit size mostly determines the size of pointers (i.e. addressable memory in instructions, C programs and on the memory bus), which is pretty much fixed

>>56176839
and we have containment boards for your kind, what's your point?

>>56176879
I'm a technology enthusiast on a technology board. There is no better board for "my kind"

>>56176841
>memory bus
it's not the 60s any more great grandma

>>56176906
Where do you expect your CPU to perform lookups?

File: 1457274399165.jpg (44KB, 544x529px) Image search: [Google]

44KB, 544x529px

>>56176921
>CPU
>2016

I thought this was the technology board, but apparently it's the rock board

>>56172824
i know that bait but good one

>>56172824
Kek'd heartily

how come 64 bit computers dont have graphics like nintendo 64??

>>56175597
i highly doubt that they're looking into 128 bit addressing because of memory limits

http://www.fnal.gov/pub/science/inquiring/questions/atoms.html

>>56176488
What trade-off that doesn't contain performance as factor are you thinking of that cause 64-bit to be the current most used architecture?

Yeah, I thought so.

>>56178748
Wait, did I fuck up? It would make sense that fully using 2^54 bits of addressable memory would require at least 2^54 places to put it. But the highest areal density of tape from wikipedia is .84 Gb/in^2, which means 2^54 is totally reachable. Holy shit what the fuck am I doing wrong.

>>56172695
no man's sky has 2^64 planets
where is your god now