What now?
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
>processor development from this point on will be 50% slower
>unless carbon nanotubes actually happen processors will hit a wall for size in 2020
What direction are computer going to head now? Are people going to be content with hardware not advancing as fast as it did? Are people now going to start shunning small size in favor of more powerful multi processor setups with better cooling? Are we going to see a bigger push to move everything to the cloud so that people can still experience computers becoming more powerful like they'd expect while maintaining the current trend in reducing size?
Here's a hint: People don't like change.
>>53645418
You move to quantum computers
You realize Intel has never been the leader I need innovation and we're only good at taking research and expanding on performance
You realize AMD has been the leader that opened up CPUs to the next level and will do it again
More cores!
>>53645418
1nm transistors have been a thing since 2012. The wall isn't there, it's in the cost-benefit curve.
>>53645418
Does making it smaller make it faster?
If not, then whats the point of making it smaller apart from muh cooling or muh phone games?
How is the cloud going to solve anything?
It is not like the servers run on magic.
>>53645418
We move back to 70's with big ass machines, as big as your mom
>>53645533
Because introducing a 100 ms delay on all operations is apparently faster than what we have now!
>>53645418
3D transistor layout.
>>53645610
Layering had been a thing for decades m8
>>53645524
not directly but yes
smaller means less heat, less heat means it can be clocked faster
also shorter tracks
The problem is that the smaller you go the more quantum effects play a role. Especially tunneling becomes a problem.
>>53645497
>You move to quantum computers
Last time I checked those won't work without serious cooling hardware that takes up a lot of space and makes a lot of noise. Might have a problem getting the average person to move to that.
>>53645533
>How is the cloud going to solve anything?
If hardware still decreases in price over time and price doesn't also hit a wall, building larger computers and larger datacenters wouldn't be significantly more expensive. People could then still see the advances in processing power as datacenters expanded without having to move to larger devices.
>>53645631
Why not improve cooling technology then?
Or improve materials so that CPUs can run at 300C
>>53645631
Almost.
Smaller gates have a shorter gate delay. There are more characteristics to a transistor than just how much power it uses.
A smaller transistor also is not necessarily going to clock higher than a larger device, regardless of waste heat generated. Transistors libraries are tailored to certain voltage and frequency ranges.
>>53645666
Charge carriers fail when they exceed certain temperature ranges. You can't fix that, its a limit of physics.
I'm still on Sandy Bridge. Whens the next great leap forward?
>>53645817
When the chairman declares it.
Anyway, don't TSMC say they're going to be producing 7nm chips in 2018?
![sshot507ecbe018924[1].jpg](https://i.imgur.com/Bmav33X.jpg "sshot507ecbe018924[1].jpg")
>>53645760
>Limits
I'm sure that 30 years ago people wouldn't have imagined that we could have terabytes of storage on something the size of a fingernail.
We'll figure something out by 2020.
>>53645760
>its a limit of physics
More like a limit of the material.
>>53645899
so physics
>>53645978
materials have different limits for different things. Limits of Physics are universal.
>>53645998
But muh ever advancing tech
>>53645998
Found the inbred
>>53646086
well, as costs go down, and things because stackable, I guess we can go forward.
>>53645899
>>53645998
Its not a limit of any one material. Its a characteristic of all matter in the universe.
Electons will escape any channel no matter how well isolated if temperature increases past the critical limit.
Go take EE 101 and get back to me before you shitpost more.
Developers will need to write well optimized decent code (which will never happen lel) and everything is going to become shit (slow games and so forth)
>>53646113
Nuh uh we just haven't found the super material to do it yet, my autism demands it.
>>53645418
Intel has already hit a wall.
No performance improvements since Haswell, just efficiency.
Are the rumors about them moving to III-V post 7nm still rumors?
>>53646204
They haven't confirmed anything yet, so they're still just rumors.
>>53646128
For a second I had hope we were going to finally return to efficient and powerful programs and UIs, then you woke me up.
I can't believe that, considering what I first used and got on the net with, the Raspberry Pi is as useless as it is.
>>53646223
I can't see it happening any time soon, given their entire infrastructure is based around Si processes. It would cost an insane amount to update all their machinery to a completely different substrate process.
The market demand isn't there for them to foot that kind of bill.
>>53645588
>Because introducing a 100 ms delay on all operations
Do you not know how a thin client/dumb terminal works?
>>53647236
Obviously you don't.
>>53645497
>Quantum computers
stop spewing bullshit you have no idea about
Most likely temporary solution we have is GaAs.
>>53645588
Operations =/= inputs
If that were true, processing power would be dependant on typing speed.
>>53647298
The software runs server side. There would just be a delay between inputs/outputs, not every operation.
No!
It will be like this:
> processors don't get more powerful anymore
> code is still resource heavy
> write light code
> java will become faster / something completely new will arise and replace java
> pajeets won't be able to keep up, because shitty windows dev.: HURR DURR I can write HELLO WORLD and it only takes 4 Gb of RAM (btw. 16GB ram meme)
> C will have a revival
> finally minimalism will come to the normal world (one program for one job (UNIX philosophy))
>>53647636
ok, back to 1980s with a terminal in every home to connect to one giant super computer . Privacy issues solved (because there's no private/personal computer anymore)
>>53647676
>>53647636
that would mean the end of /v/!
>>53647636
>POWER OF TACHYONS!
>>53645631
It is also easier to burn a twig than it is to burn a chunk of wood.
>>53647662
>C
>revival
Top fucking kek
>let's use the most insane language ever created even though it literally has 0 advantage compared to its contemporaries, let alone modern languages
>>53647723
at least I made someone laugh!
C was the wrong thing.
But let's be honest, Java IS bloated (a GUI application takes an uncomfortable long amount of time to load even if it's small. Many old computers reacted (almost) INSTANTLY whenever a key was pressed, java can't do this because of BLOAT + stupid garbage collection (eats up ram)).
Everything is getting bloated, not just java! Why can't we have computers that do things (almost) instantly (I mean not instantly, but so fast that your finger hasn't even left the key before the task is completed). In the past this was possible!
>>53647723
>>53647796
enjoy android java bloat lag !!!! (<-- see what I did there, multiple memes in one line!)
All my technology does literally everything I could ever want.
>>53647676
Not saying it's a good idea, but if the situation could move toward that then it's better to start planning for how to deal with that now than later.
>>53647700
Not really, you just have the game clients and server running in the same datacenter.
>>53645519
Sauce?
>>53647796
Java is exactly the wrong thing in the same way as C is the wrong thing: both are powerless, bloated, inconvenient languages that produce only dangerous software. In the case of C because it's completely unverified at compile time, in the case of java because there is no safe JVM available.
>>53645418
Buy used CPUs from server selloffs once X99 chips hit the market. (currently we have 8core sandybridge chips going for 70 bucks) Keep that system until optics take over.
It's 2016: has graphene changed the way we look at CPU's yet as we've been promised every other year
We're at a point of good enough. 3rd worlders still need to get their flagship tracking devices though.
I see custom setups for those who require raw power like the turbine mac pro. The rest of us will use low powered devices for mundane every day shit.
>>53648087
Get 2 E5-2670s and a dualsocket 2011 board and 128gb ECC memory and forget about upgrading for the next 3 years.
Um.. http://arstechnica.com/gadgets/2015/07/ibm-unveils-industrys-first-7nm-chip-moving-beyond-silicon/
>>53648570
it's not a rainbow anymore, into the trash it goes
>>53648801
Its because the structure size is so small now that it no longer diffracts visible light
>>53647992
You're a functionalfag aren't you?
Kill yourself.
>>53645978
>>53645899
>>53645760
So come up with a carrier that has different temperature properties and will not stop signal transmission after reaching a certain temperature - or only change at a significantly higher temperature.
Less power and resource consumption for a better carbon-neutral planet. Less pollution. Smaller hardware that takes up less energy and performs faster.
I read somewhere that one of the reasons Banana Computers makes stuff smaller is because of environmental impact. Less consumption, plus it's what consumers think is neat.
New thesis are about building microprocessors linking components vertically as it reduces space increases computational power and saves energy
>>53645865
>people wouldn't have imagined that we could have terabytes of storage on something the size of a fingernail
Maybe the average consumer, but if you crack open any old magazine there wasn't a ton of pessimism about the coming years like there is now.
due to conservation of space/time there's no real reason 'quantum' computers would be any faster than conventional methods. it's a fucking meme.
>>53649040
>because of environmental impact
inb4 old computers are sad
>>53648950
Forgot about that one. http://www.purdue.edu/newsroom/research/2012/120219KlimeckAtom.html
>>53647805
your meme formatting is harder to read, needs documentation now
>>53649424
>due to conservation of space/time
literally a meme
Abandonment of von Neumann in favor of graph reduction oriented architectures
>>53649157
>if you crack open any old magazine there wasn't a ton of pessimism about the coming years like there is now.
solid source ;^)
cant they just 3d print the cpus instead??
>>53645494
>Useless words: the post
>Apps are more and more like desktop programs
>Eventually there are just terminals and you plug in your device to access a desktop environment much like continuum in ubuntu or that thing Windows is doing which I cannot be fucked to Google
>Mobile devices become more and more mobile to the point that wear-ables are the power source for these things. SoC's include nand flash and require very very little power.
Maybe something to do with implants for health tracking and modification?
>>53645418
More specialization, more ASICs, more special deep learning hardware.
>>53645418
>Are we going to see a bigger push to move everything to the cloud
Yes.
If you want to predict the future with >50% accuracy, just assume that horrible things (and in technology, things that compromise consumer privacy and ability to modify their own machines to the benefit of large companies.) will happen.
