It's over. Computer's are finished. No more computational

>>55904333
The bottleneck is now on the shoulders of programmers.

The future of computational improvement lies in massive parallelisation. It's the fault of lazy programmers who are still thinking in a single-threaded way that runs everything on top of a fucking JIT VM.

>>55904333
very healthy

>ditch silicon
>processors can now go above 100ghz

File: 1ca1f8ea-98a7-4d92-9125-2331bc33da0b.png (143KB, 300x300px) Image search: [Google]

143KB, 300x300px

>transistor size determines performance of processors
>3D processors will never be made

>>55904333
>No more computational progression.
Why don't you kill yourself?

It's scary that there isn't a real alternative.

>>55904398
No such thing

>>55904403
Only increase in physical volume, that's not advancements.

>>55904417
Show ONE thing that can continue this trend.

And boys, even if we find a new material it wont' last long, perhaps a decade before it reaches atomic level again and then we'll be completely finished.

Have fun having the same computers forever.

Huh, guess it's time to go back to banging rocks together then.

>>55904333
designers thrive with limitations.

faster we get to 5nm better shit we get.

File: 0e3.png (187KB, 600x339px) Image search: [Google]

187KB, 600x339px

>2021 arrives
>get a ThinkPad workstation
>never have to buy a PC again

>Matrix-like level simulation will never be achieved

I'll get me some rope.

File: 1367458589113.png (4KB, 354x353px) Image search: [Google]

4KB, 354x353px

>>55904333
Almost nailed it, but there is still some stuff we haven't tried. Mainly novel processor and memory typologies, like processor fabrics that can access adjacent core's memory, but not any sort of global memory. But yeah, the Moore had a good run, but it's pretty much over.

>>55904398
Too expensive to use other types of semiconductors.

>>55904403
Although this could help with the issue that we can't move information across a modern chip in a single clock because of the speed of light being an issue, we have a hard enough time cooling mostly 2D architectures. 3D will only get us a few layers. Besides, do you have any idea how many layers are on even 2D processors? Like 30.

>>55904417
Butthurt? Complain to the fucking UNIVERSE about PHYSICS.

>>55904455
>No such thing

http://wccftech.com/graphene-transistors-427-ghz/

http://www.extremetech.com/extreme/175727-ibm-builds-graphene-chip-thats-10000-times-faster-using-standard-cmos-processes

>>55904641
Dear UNIVERSE,

>>55904681
>an astonishing 427 Ghz!
Just hype.

It's like discovering photons and believing we can travel at the speed of light.

>>55904694
NIGGER

>>55904372
Plenty of room for improvement at least.

We can all go back to writing C instead of fucking java.

Silver lining?

>>55904681
just read the article.
The most retarded article I've ever read.
You can even see it in the comments.

Quantum Computers.

The use cases you are describing are but the tip of the iceberg in the future. If it can run calculations, it can do anything.

>>55904882
never read prajeet techblog comments unless you need to lose your mind with a quickness. or read them crying about I AM NOT GOOD WITH COMPUTER I WANT TO TOUCH TITY SO BETIFEL SUCC ME from thirsty ass indians and pakis.

>>55904641
>typologies
>making a typo that is literally 'typo'
pure pottery

Ray Kurzweil has said that Silicon is the 5th paradigm and there will be a few more paradigms until the singularity. I think we have atleast another 10 years of solid progress for silicone before we hit a definite wall. We dont have to keep shrinking transistors to have progress. Theres other techniques that could be utlized once we hit 5 nm.


Most of the promising long term replacements for Silicon unfortunatrly use expensive or out of reach materials like Graphene or Carbon Nanotubes.

But once we are able to create graphene and CNTs in sufficient yields and quality, we will have better computer that will be thousands of times better than silicone chips.

>>55905174
>Ray Kurzweil has said
opinion discarded

>>55905202
Hur durr edgy

>AI meme won't happen

It's been nice daydreaming about it, lads.

>>55904333
All x86 CPUs newer than Core2 and Bulldozer are botnet anyway

>>55905332
k

Does this mean my t420 will last me the rest of my life? I'm okay with this

>>55904372
>It's the fault of lazy programmers who are still thinking in a single-threaded way
As long as later operations need results from previous operations, that's always how it's going to be. It's not necessarily the programmer's fault.

Look, fucktards. You know that thing that holds your ears apart? It's a fucking
>computer
And it stomps your Stinkpad in the ground. Why?
>Parallelization
Hundreds of billions of neurons. We've barely even scratched the surface of what is computationally achievable.

File: zisblatts'colon_cleanse_6million.jpg (64KB, 281x537px) Image search: [Google]

64KB, 281x537px

>>55904681
whatever happened to ?
http://www.wired.com/2003/09/diamond/

>>55905951
in theory yes, but we reached physical limitation, sry.

>>55904333
>10 nm – 2016
shiggy
>7 nm – ~2018
diggy

>>55905004
>If it can run calculations, it can do anything.
the height of stupidity
quantum computers require quantum algorithms, which are not always going to be faster to compute than normal algorithms
the most computational intensive task most computers have to perform is video playback, and sometimes rendering graphics for video games

we might see quantum co-processors at best if there are any worthwhile quantum algorithms in those two areas

>>55906140
Video playback is literally nothing compared to rendering vidya graphics.

>>55906189
I didn't say it was, just noting that those two areas are the most computationally expensive ones.
Hence why you see a lot of multimedia instructions being added to CPUs time and time again.

>>55904333
Honestly it just seems they make everything now bloated when technology gets faster.
Im taking about programs and Web pages.

Imagine if they used something that required as much memory/cpu as windblows xp or dial up Internet pages? We could do things crazy fast.

But no let's make an os that requires 4gig of ram and 500mb sized webpages

>>55904333
Umm. Trinary?
Also, why the fuck don't we have it yet.

How come no one mentiones bio computers?

https://en.m.wikipedia.org/wiki/Biological_computing

>>55906140
Quantum computers/chips will be amazing for AI, 'nuff said.

>>55906400
Oh yes, I need to run a highly sophisticated fucking AI on my laptop that I use for watching YouTube videos and pornography.

honest question here

why dont they make the CPU just bigger?

>>55906450
bigger = hotter
eventually though, once we hit the transistor size limit, we will be making bigger processors

>>55906450
>>55906472
If we can pack down what we could and more for 22nm in 14nm let's say, why don't we pack even more than what we could fit in 14nm in 22nm if that makes sense?

>>55906425
You don't need to, but at some point in the not-distant-enough future it's going to get shoved down your throat anyway.

>>55906425
Indeed. you have have pre watch things and give it a rating based on your preferences, vastly improving your masturbation sessions by filtering out click bait.

>>55906511
>Indeed. you have have pre watch things and give it a rating based on your preferences
YouTube already does this.

File: they-dont-think-it-be-like-it-is-but-it-do[1].jpg (154KB, 600x826px) Image search: [Google]

154KB, 600x826px

>>55906503

Diamond and graphene processors are already well into development, so no not only is it not the end of computational progress dare I say its the beginning of monumental leaps and bounds.

in the meantime there's actually still quite alot of architectural tweaks and designs left in silicon so do not worry you'll have more fanboy bullshit to spew in the meantime i assure you.

But yeah the easy path of 'just shrink the die again' is coming to a close. And it's about damned time.

>>55906503
I don't understand what you're trying to say.

>>55906572
>And it's about damned time
Yes, I am also tired of CPUs just werking, I want them to fuck the architecture up.

>>55906586
IF WE CAN PACK DOWN LET'S SAY A BILLION TRANSISTORS IN 22NM AND 1.5 BILLION IN 14NM THEN WHY DON'T WE PACK 2 BILLION IN 22NM

>>55904333
Good. It's not quite breaking free from the shackles of technology, but it's better than blindly Moore's Law-ing toward all-powerful AI gods controlling every aspect of human society, assuming they don't just wipe us out like bugs.

>>55904681
More like Graphmeme

It's the hottest shit in STEM and no one can figure out how to use it.

>>55906597
okay so you are fucking retarded, thank you for clearing this up

quick question: what do you think those nanometer measurements represent?

>>55906450
If you make them larger, you will hit some issues about the speed of light not being able to travel through the all CPU in on clock cycle
Information is not transferred instantaneously

>>55906618
calling bullshit on this one

File: 1469497371955.gif (318KB, 500x500px) Image search: [Google]

318KB, 500x500px

>>55906400
Quantum non-deterministic computers aren't any more capable than deterministic computers, they just solve certain problems faster. At most it will be akin to a GPU. Never will you run general purpose computing on a quantum machine.

>>55905993
https://www.google.com/patents/US6858080
https://www.google.com/patents/US5635258

>>55906612
Distance between identical components on the fabricated chip

>>55904333
>long-run investment in R&D for one specific technology has declining marginal rate of return
who'da thunkit?

>>55906450
Because a bigger processor is more expensive to manufacture and has lower IPC which matters to those who play severely CPU crippled vydias that can only run on 1 CPU thread efficiently.

We actually already have a 24 physical core Xeon processor that can run at 2.2 GHz when under maximum load. Has about 3X the performance as a desktop i7 with only a TDP of 165W. Did I mention it costs $7,000+?

http://ark.intel.com/products/93790/Intel-Xeon-Processor-E7-8890-v4-60M-Cache-2_20-GHz

However like I mentioned these fuckers are god dam expensive. So much that those building servers or supercomputers opt for cheaper 8-core Xeons to put in parallel.

>>55904333
>they're useless for anything other than encryption and quantum physics
>>hurr durr i can read stuff of which i barely understand anything and then start shitposting about it leaving the impression i know what the fuck i'm saying
>>t. Dr. summerfag, future mcdonald employee of the month

>>55906618
>Biological_computing
yea but it doesnt matter if it arrives at its destination according to our frame of reference as long as its time happens relatively the same in its own space-time, and thats not even accounting for mass or gravity

>>55906658
in case you haven't googled it yet, nm measurements within the context of processors is the size of the transistors

>>55904333
Although I disagree, I don't feel the need to argue and only came into this thread to examine the breasts of that woman and blog about it.

Thank you and no thank you at the same time.

File: 1465751571525.jpg (63KB, 450x562px) Image search: [Google]

63KB, 450x562px

>>55906629
Light travels at 3.0e8[m/s] IN VACUUM

Let's say you are running a 4GHz machine.

1/(4e9 Hz) = 0.25e-9 s

0.25e-9[s] * 3.0e8[m/s] = 0.075[m]

That means at most your in vacuum CPU can move a information 7.5[cm] per tick. Now, CPUs are built in silicon, and such light propagates slower. Also traces don't go the way the crow flies. These are back of the envelope calculations, but show that die sizes are and clock speeds are truly limited by how fast light travels.

It's actually worse than anon said. Presently modern CPUs take more than a single clock tick to move data long distances. Optical buses may help alleviate that a tiny bit, but there would have to be substantial room for optical couplers and decouplers. Moreover, optical buses are only useful for long distance buses, not short ones.

>>55906593
blow me edgelord...it is about damned time. best work is always done with a difficult dilemma.

this will do nothing but spur advancements as such things usually tend to do. This is not the first time jackasses were wailing about how the cpu cannot progress any further no matter what. This shit happened like 1/2 a dozen times already and every time a way to overcome has been found and implemented.

Now we have the opportunity to explore new things. things are going to get really fucking exciting in the next 5-10 years. You wont be disappointed.

Except for your failure to troll...that will always disappoint.

>>55904768
>parallelization in C
god damn i might just wanna kill myself from reading that phrase alone
handling mutexes/threads in C is a total fucking nightmare

File: moar.jpg (166KB, 1070x1040px) Image search: [Google]

166KB, 1070x1040px

Is there any reason we cant just make processors bigger?

>>55906757
Yes, if you could read the fucking thread.

>>55906750
lol wut. it's the only sane language to do that stuff in.

>>55905951
A brain is much different than a conventional computer.
To give you an idea, a conventional computer requires memory (in which instructions and stored data lies) and a processor (an set of transistors organized as logic gates, designed to be controlled by a certain instruction set) which does the computation.
Whereas in a brain, the neurons do both the memory and the calculation simultaneously. Imagine having no HDD and your processor doing all the work without memory or instructions, and you have a brain.

>>55906750
care to expand a bit on it? only did mutex/threads in java, though i can't see why it'd be such a difference in C

>>55906738
It's a fucking miracle with the huge ass instruction sets we have now that processors work as well as they do.
I have no doubt that we will continue to make computational progress, but switching from transistor shrinking to architecture overhauls is going to be extremely painful.

>>55906771

>read thread
>you're just getting butthurt and calling people stupid over and over

Real fuckin educational. Go jump off a building, faggot.

>>55906757
see >>55906675

Nigga, do you want to pay 7K+ for a fucking processor?

>>55906750
>handling mutexes/threads in C is a total fucking nightmare
You literally lock and unlock a fucking thing, how dumb are you?

>>55906816
literally explained 3 posts up from your stupid image macro you fucking faggot, kill yourself

Why dont these idiots just add more cores and makes the CPUs bigger

ffs

File: elN3kho.jpg (9KB, 576x1024px) Image search: [Google]

9KB, 576x1024px

>>55906878
please don't bait with the others.

>>55906773
if by sane you mean it's gives better performance than things like java, then yes

>>55906786
it's not too much different but on top of all the other errors you can run into that the language manages for you in something like Java it can be very difficult

>>55906836
any program getting reasonable parallelization will generally spin off more than 2 or 3 fucking threads, managing them properly so they don't deadlock can be nontrival in any program solving a real problem

>>55906922
>muh triple threads!
So how the fuck do other languages make it any better, besides adding a bunch of retarded constraints that you can add with C?

>>55906940
i usually use c# for this kind of stuff
so I guess my argument boils down to the fact that it's more built in to the language than it is with something like C

>>55906750
use an existing thread pool implementation like glib/ck/openmp has, you don't have to touch that shit just because it's C

>>55906597

Nigga you trolling or what?

NM means the size of the gate grid. Each gate grid can fit 4x4 transistors in it.

Common knowledge

>>55904455
>Only increase in physical volume, that's not advancements.
AFAIU, advancements stopped a while ago already. the latest processors are not much better than those made in ~2008-2010... except they use more power to get more computational power, and minor improvements in the designs to get a bit more juice

Is the singularity a meme?

Why we can't make cpus bigger?

File: le rabbit man.png (24KB, 739x675px) Image search: [Google]

24KB, 739x675px

>>55904580
>>55906425
>>55905246

>run AI on quantum computer
>AI replaces "lazy programmers" for the stuff that has to use traditional algorithms

>55904372

>>55906140

>>55904641
>Too expensive to use other types of semiconductors.

what if everyone gets a dumb screen with wifi and access huge mainframes?

>>55906608
>tfw God is watching over us

>>55907423
shhhh

They're listening. That's probably what'll happen in the future.

>>55904333

who is this semon demon?

>>55907362
no, it is not.

People saying it's going to happen within a hundred years are fucking kidding themselves though.

>>55907516
This. Google returns a dead-end.

>>55904333
I want to impregnate those tits

Also, Non-Si transistors

>>55906897
shut up

>>55905951
>hundreds of billions

>>55906609
It does have some really impressive properties and I'm still hoping we find a way to use it and mass produce it efficiently and cheaply in the near future.

File: Chrisautismmeme.jpg (436KB, 810x810px) Image search: [Google]

436KB, 810x810px

>>55905951
>>55907802
>>55906780

Why hasn't anyone mentioned expanding the bus? I know x64 to this day isn't fully utilized for some goddamned reason. But what's preventing a jump to x128 other than the obvious cost and horrific adoption rate?

>>55907967
Fuck it, why not a jump to x65536 other than the obvious cost and horrific adoption rate?

>>55906737
well, thanks for explaining

>>55904333
>implying this is a bad thing
Now we are done with making the design smaller, we can focus on other aspects.

>>55905332
>botnet
Nothing to hide, nothing to fear, weed is decriminalized

>>55904372
fuck that, algorithms are going to design the computers of the future

>>55904333
That's what they said in the late 90s, you're an idiot.

>>55906140
Dafuq do you know about quantum algorithms that haven't even been designed yet, anon? It's all about implementation and architecture. Eventually quantum computers will surpass our silicone processors in every way imaginable.

This is inevitable, deal with it.

>>55908019
Still a more viable option than quantum computers.

ITT: Pseudo-computer scientists/engineers
Some of you are all right though.

>>55909064
>don't go to lab tomorrow.

>>55904705
>set Dwarf Fortress framerate to be uncapped
>five seconds in you've lost to a tantrum spiral

>>55906773
I think you mean Rust

>>55904333
Holy shit, yes
Bring on the technological stagnation. Maybe we can start actually focusing on repairing the software and shit that we have now without having a bunch of idiots reinvent the wheel every six months

>>55908613
yeah but they still take it if they catch u with it where i live

>>55904333
>"computer's"
kill yourself

>>55906737
sorry for being retarded but i dont understand the role light plays in a processor. i thought it was just electricity being routed through the gates, i don't understand where light comes into the equation. clearly i dont know how a processor works

>>55909000
>do you know about quantum algorithms that haven't even been designed yet
>Eventually quantum computers will surpass our silicone processors in every way imaginable.
it seems like you know about some new algorithms
why don't you tell me about them, you stupid piece of shit
also tell me about how quantum computers will surpass regular computers when you don't have the space or power to fucking cool them down enough to get quantum effects

stupid retarded faggot
don't reply to my post

>>55909280
rude
>>>/b/

>>55906750
>what is OpenMP

>>55909294
I just told you not to reply, idiot
post more nonsense like that again and I'll shit down your throat again

File: 1467500233794.jpg (204KB, 760x1024px) Image search: [Google]

204KB, 760x1024px

>>55904333
there's photonics and chems and carbon and lions and bears and shit

this ride is just getting started


buckle up buckaroo

>>55904372
>shoulders of programmers.
LOL then we're fucked

programmers will oversaturate as talent continues to spread thinner and thinner

just kidding, i know it's always just been a few bright bulbs doing most of the heavy lifting and the rest of the devs are just filler like digital burger flippers

>>55909383
we just need the few bright ones to make algorithms/frameworks/designs that the rest of the programmers can then abuse and misuse

no different than what is happening right now

>>55904455
>have fun having the same computers forever
And whats the fucking problem with that?
Finally we will be able to end this "keeping up with the jonses" shitfest of a pc and console community, there'll be nothing left to enjoy but the content. Fucking brilliant.
Fuck off anon, I welcome the future.

>>55909443
>implying we won't still have a keeping up with the joneses situation with hardware architecture
your dream will never come true so long as a monopoly does not exist

>>55909464
It's not a dream faggot, it's a possibility.
I don't stick my hand up my ass like you do and caterwaul at a Taiwanese macrame appraisal forum when I see a possibility of the future I don't like.

>>55906312
Yep, you can put a whole Forth OS and goodies into a couple K.

We will have to remember the old ways.

>>55909496
>It's not a dream faggot, it's a possibility.
You blowing your worthless brains out is also a possibility, but I'd still call it a dream of mine

File: anime 89234.jpg (82KB, 1280x720px) Image search: [Google]

82KB, 1280x720px

Does this mean AMD will become competitive by 2026?

>>55909541
Amd won't exist in 2026

>>55909565
as long as anti-monopoly laws still exist, they will

>>55909595
TRUMP FOR ahh fuck it nevermind

File: 1469331520682.jpg (159KB, 810x1080px) Image search: [Google]

159KB, 810x1080px

>>55909277
well, for starter, information can't go faster than light, so what I showed was the absolute longest distance you could go in a modern clock tick.

secondly, I'm using light in a broad sense as an electromagnetic wave phenomenon. Even though visible light isn't used to transmit signals on most chips, the electromagnetic fields that are used can only propagate so fast. We use metal traces to route signals around. It turns out that when you confine an electromagnetic wave inside a trace, it slows down. Thus signals actually move slower than the speed of light just because of the geometry. Moreover, materials universally slow light down compared to vacuum.

TL;DR because physics

I'd like to compute over dem titties

>>55904641
>Too expensive to use other types of semiconductors.
Everything gets cheaper at scale

>>55906737
Familia, that means that each individual core can only be 7.5cm long.
Which turns out to be pretty fuckin' big.
The entire CPU need not be 7.5cm, but register-register in the datapath must be 7.5cm.
This distance is usually maximized between the ALU and the register cache.

>>55909153
I think you mean erlang, faggot.

>>55909800
>Information can't go faster than light
hurrrrrr
Of course it CAN, it simply can't be accelerated from less than the speed of light, to beyond the speed of light.
We could harness those FTL particles in a computer system.

>>55906737
>>55910093
I also hate to tell you that your hypothesis is wrong, considering people have made massive CPUs before.
http://www.megaprocessor.com/

>>55904455
>Have fun having the same computers forever.
I already have the same computers forever, kill yourself you enourmous piece of shit. This thread is more cancerous than desktop threads. I need to puke.

>>55904398
Light has a limitation of around 10Ghz

>>55904403
>3D processors will never be made
You know that we have multiple layers of silicon? They ain't just one flat 22nm surface with transistors on it, they grow in height too with every major release.

>>55906375
It hurts my brain to think about how somebody could be this much of an idiot

>>55904403
3D processors are already being made. Everything below 28nm is 3D.

>>55904333
What kind of problems are we trying to solve with more multiply and accumulate?

>>55910445
It's about improved power efficiency, more calculations for your watts.
More calculations for your $ as well would be nice, but since watts cost money you might as well go with that.

When warp drive becomes invented, we'll use that tech on our processors and say "fug u speed of light" and just like create the singularity and shit.

>what is indium gallium arsenide
/g/ autists not knowing shit as usual.

>>55904768
they don't teach C in liberal arts though

They're not. Someone builds a faster computer very fucking year and several novel prototypes.

I guess your gaymer consumer hardware is all that exists though

>hurr intel won't make a 3d i7 computing is dead
No the consumer market is dumb and needs compatibility more than capability

>>55904333
>/g/ is dead
>also technology is dead
welp at least there was some solace in knowing the demise was inevitable

>>55906836
You have no idea on how hard parallelization is. Remember: there's a reason CSP, Actors, STM etc. exist.

File: 1448065236139.gif (871KB, 245x230px) Image search: [Google]

871KB, 245x230px

>>55904564
to bad all those electrical components that were designed over 80 years ago still breaks after ~5 years of usage and forces you to buy a new one anyway

but you wont really have to worry about that since the pc market will already have copied the smartphone industry and started to design their hardware to break more easily so that people will continue to buy replacements every year to compensate for the lack of upgrade needs

>>55910195
Senpai it's not even comparable to modern integrated circuits. It's a 16bit processor running at 20kHz. The reason it's so slow is because it's so big

The reason it's so big was for teaching, not JUST to make it big

>>55904333
fucking finally
I won't have to upgrade my gpu

>>55905731
The electron waves just need to go back in time is all.

>>55910093
>implying an electron moves even close to light speed in silicon

>>55904333
>he actually uses computers
Fkin nerd.

>>55911032
>implying that electrons move at all
>not knowing what electric field is

>>55904681
>Let me put it this way. Remember that old calculator lying around in your desk drawer somewhere? Thats how our current generation will look compared to a Graphene Transistor based chip, and even then i am being generous.

>>55909800
>information can't travel faster than the speed of light
But that's where you're wrong anon. It can be transferred instantly. The medium of information that we currently employ can't travel faster than light.

>>55911121
They move but not in the way most people imagine

>techfags rely on SCIENTISTS to advance their codemonkey lives

We scientists are superior .

>>55911524
Has anyone ever disputed that?
AFAIK, the hierarchy has always been
>Theoretical Science
>Practical Science
>Engineering
>Implementation
>Utilisation
>(Maintenance)

>>55911619
>(Maintenance)
the tragedy of the human race, everybody wants to create but nobody want to do maintenance

>>55904333
>Computer's are finished.
Does that mean you're finally leaving

>>55910920
This.

Finally progress can halt for a while and we'll be able to enjoy what we have.

>>55904372

>The future of computational improvement lies in massive parallelisation.

Correct.


>lazy programmers who are still thinking in a single-threaded way that runs everything on top of a fucking JIT VM.

You got it wrong..

The very best languages for this task use a VM..

BEAM > JVM, baby

>>55912127

>The very best languages for this task use a VM..

The very best languages for this task use multiple VMs:

LING > BEAM > JVM

The LING VM is a VM which runs directly on the Xen VM hypervisor - without any OS needed:

Hardware --> Xen --> LING --> Erlang/Elixir

This makes is processes super fast and extremely lightweight.


Where is your god now?


>http://erlangonxen.org/
>https://medium.com/@kansi/bare-metal-erlang-hello-world-9cea0ea702ea

>>55906940
Love the inter-board joke

I have no idea what you faggots are talking about but I don't are, fuck off.

I want my quantum dragon dildo already you fucking Terry Davis wannabe.

FUCK OFF CIA NIGGERS

No one really cares, at least in regards to desktops. It's not the 90s anymore where every update in CPUs had huge, clearly-visible, day-and-night improvements. For most people, the upgrades now days are simply redundant.

>>55904333
>Computer's are finished
he wrote on his computer

I don't get it. I know we're making transistors smaller so we can fit more in the cpu. When we hit the 5nm wall can't we just make the CPUs themselves a bit bigger? Or just layer them? Its not like they're too big as it is.

>>55913133
Layer, to a degree yes (muh heat), but not make them bigger. Read the thread. You're like the 100th person to make this original intuition.

>>55904333
Sauce?

>>55904333
>Don't mind me, I'm just here for the thumbnail

>>55910192
>>55911283
this fundamental principle of the universe is wrong because I believe in magic
>hurr durr "Quantum" buzzword
come back when you have a real understanding of quantum particle physics

>>55904333
>what is powerpc architecture

>>55913668
>hurr durr I don't follow science

Shut up faggot.

http://phys.org/news/2015-09-physicists-distance-quantum-teleportation.html

>>55906634
Except quantum computers are great for decision making since they can answer non-binary questions, hence improving AI dramatically

>>55904333
Fix your roadmap. 14nm was 2015, 10nm is next year, and 7nm on are no longer planned.

>all these replies
>"b-but CPU architecture"
How dense can you be?

The miracle of computers was that it was exponentially growing in power, as time progressed we'd have crazy technologies in the future.

The computing power has increased million times (if not more) in power when it was first introduced.

Optimizing architecture simply means a slight increase in efficiency of the EXISTING number of transistors, the difference will be almost zero compared to creating smaller transistors.

>b-but quantum computers
Jesus fucking christ, stop reading sensationalist articles, literally anyone involved in developing Quantum Computers have repeatadly stated that Q.Computers can and will never replace classical computation because of the radically different nature.
>but if it can solve problems the it-
STFU YOU LITERALLY UNINTELLIGENT MONKEY THAT'S NOT HOW PHYSICS WORK

>ah at last we will create better software
Screw that, fuck that.
It's like being stuck in bronze age and refusing to further your technology and say "at least we'll make some quality bronze items"

Just kill yoursselves already

File: photonic-chip-6_750x500.jpg (36KB, 750x500px) Image search: [Google]

36KB, 750x500px

Screw quantum, that's about as meme as they come and there's no way you can rake in continious profits from something like that.
It's not going to happen.
Next step is going to be the Photonic CPU
They've been pumping money into them for a long time and they're actually getting results.
First functioning dual core CPU was built about 8 months ago.

http://news.berkeley.edu/2015/12/23/electronic-photonic-microprocessor-chip/

>Chip had a bandwidth density of 300 gigabits per second per square millimeter, about 10 to 50 times greater than packaged electrical-only microprocessors currently on the market.

Also they don't consume shit when it comes to power.

>Using only 1.3 picojoules per bit, equivalent to consuming 1.3 watts of power to transmit a terabit of data per second.
>In the experiments, the data was sent to a receiver 10 meters away and back.
>The advantage with optical is that with the same amount of power, you can go a few centimeters, a few meters or a few kilometers,”
>For high-speed electrical links, 1 meter is about the limit before you need repeaters to regenerate the electrical signal, and that quickly increases the amount of power needed.
>For an electrical signal to travel 1 kilometer, you’d need thousands of picojoules for each bit.”

This is the future tech right here, forget quantum.

>>55906609
Transparent contacts with it for solar cells haven't reached a mature stage because of manufacture limitations but the process is reasonably robust if you have daddy government's cash student and grad students with nothing better to do.

File: pijobf4e8h.jpg (133KB, 550x500px) Image search: [Google]

133KB, 550x500px

>>55906375

File: 1458308375940.gif (537KB, 480x270px) Image search: [Google]

537KB, 480x270px

>>55904333

>>55914043
>Except quancomputertum computers are great for decision making since they can answer non-binary questions, hence improving AI dramatically

That's not at all how quantum computers work, and AI with classical computers already does that. Usual classicifcation neural network models spit out answers as probabilities and not certain states, already (although it's usual to just use the most probable one as the result). Remember that there is nothing that can be calculated by a quantum computer that can't also be calculated by a classical computer. In fact, quantum algorithms are usually run with a simulation of a quantum computer on a classical computer.

AI always comes down to linear algebra in the end. It's very parallelizable, but quantum algorithms cannot improve on classical ones (which can parallelize just as easily) for matrix-vector multiplication except in certain special cases that don't come up much.

>>55915551
Quantum Computers will become just another part of the PC.

>>55907362
The basic idea is not, technological feedback and development has actually been increasing for some time in general (think about when a lot of the things you use were invented). Someone invents something, turns out it can be used to make some other thing, which can be used to make even more shit, and so on.

At any rate I sure hope it's coming in my lifetime, because dammit I want to be a spacefaring cyborg.

>>55904580
>Implying it hasn't been achieved already

>>55909145
>set DF framerate cap to 10000
>still runs at 2 fps on my laptop

Are they ever going to release source? I would love to see what savage Tarn has been up to.

>>55904333
Time to optimize our codes then...

>>55907404
>implying we can't

>>55914309
I imagine they run cooler too, right?

File: RingResonator_6.gif (2MB, 1000x700px) Image search: [Google]

2MB, 1000x700px

optical circuits and ternary logic will get rid of our stupid and ugly binary thinking

let the new age begin already

File: Welcome__welcome_to_City_17.jpg (692KB, 1280x1707px) Image search: [Google]

692KB, 1280x1707px

>>55904333

Who is this muddy jizzcaddy?

>>55904580
>literally blue-pill

File: 1604079.jpg (265KB, 786x862px) Image search: [Google]

265KB, 786x862px

Would there be any point in changing our architecture from the typical binary computer model to something like decimal, or even hexadecimal?

Having 10 points of data per 'bit' (0 to 9) instead of 2 points (0 or 1) would surely give us some massive gains in data throughput, right?

>>55920811
> Being this ignorant