Now that Moore's Law is super fucking dead, what do you

>cant advance cpu anymore
>focus on lowering production costs
>apply free market
>cpus get cheaper
>programmers cant act like resources are infinite
>have to optimize software to remain competitive

future is bright anons
im not glad moore is dead but im glad hes gone

>>57399052
>programmers cant act like resources are infinite
This so fucking hard. Garbage software was literally the worst thing about Moore's Law.

Cheaper equipment will enable:

Low latency distributed systems
Infinitely parallel serverless computing

Moore's law is dead but no longer relevant anyway

>>57399052
>apply free market

>market dominated by a couple companies cooperating rather than competing
>remotely free

>>57399081
>>57399052
>>57399017
>Moores law only applies to electronics
I wish it also applies to software optimization. I want to play Battlefield 1 on muh intel atom.

Have we hit moores law on SSD's and HDD's?

>>57399766
no

>>57399017
>Now that Moore's Law is super fucking dead,
http://newscenter.lbl.gov/2016/10/06/smallest-transistor-1-nm-gate/
https://www.extremetech.com/extreme/185027-the-vacuum-tube-strikes-back-nasas-tiny-460ghz-vacuum-transistor-that-could-one-day-replace-silicon-fets

>>57399017
>>57399879
old news

I've visited labs that have single atom transistors, but they are having a hard time getting smaller.

That said 5 atom transistors are basically required if you don't want a miss placed electron to screw everything up, and electrons get miss place all the freaking time at room temperature.

My stable modeling limits drop off sharply at 15 nm, which is why Intel's 14 nm makes me nervous as their yield rate is pushing cost limits and likely adding more checks to deal with the higher error rates. Still I doubt they would release it if they hadn't found some solution, but 14 nm are already noticeable unstable in ways the average person would not notice.

But yes on a cunsumer level Moore's Law is dead.

>>57400290
instead of decreasing the transistor size, you could also increase the frequency of the circuits to increase performance. Silicon does not really allow this but other materials might.

Also quantum computing

>>57400754
Quantum computing is shit for consumer applications and doesn't speed up most algorithms at all.

>>57400754
Yes, but it much harder and simpler than that.
With every computing system you have common values, usually something like

Unit size
Speed
Energy density/Thermal dissipation
Specialization/Optimization
Error rate

We are frighteningly close to the physical limit already. Every time you mess with some part the whole thing changes.

As for quantum computing is basically a dream. As the error rate is over 50%, some of the high profile "success" stories actually give 99% error rates if you read the fine print. So they give the answer insanely fast, but it is wrong most of the time. So it is little more then gloried random guessing. The goal of reducing the error rate is funny because as they lower the error rate they lose in the other categories.

You can actually use a thermodynamically model with flops as an energy action unit, as flipping a bit can be defined as the work of moving the electrons when given a fixed volume and avalible energy. This gives you a theoretical limit constraint by physic, but the piratical value is less inpressive as the outside environment messes with things a lot more then people care to admit.

>>57399017
Pretty much what this anon here said: >>57399052, except add everything being pushed to the cloud because people want their mobile devices to continue improving. Computational power on cloud based servers would become like a utility where you get charged for how much you use.

>>57401741
>except add everything being pushed to the cloud
>Computational power on cloud based servers would become like a utility where you get charged for how much you use.
That's the thing I fear most, but as things are now it's pretty confirmed to be where we're headed to.

switch from electrons to photons. photonics gonna be the future. we just have to find a good transistor material

>>57399137
this
there's no free market

>>57402535
Photonics is a meme, spintronics is the real future.

>>57402481
Yeah, people don't care much anymore about their privacy in regards to things like that and would be happy letting companies datamine everything they do with their computer if it meant they could have a smartphone with comparable processing power to a desktop for cheap. Not to mention that with the way software and DRM are going now days we will likely see all the negatives of running everything in the cloud start apply to regular computers anyways. One of the main things that worry me if CPUs stop advancing in such a situation though is that companies might stop upgrading and surplussing old servers and instead manufacturers might improve ways of recycling old equipment, which would kill off what could be an inexpensive way to own real hardware if everything got pushed to the cloud.

Sigmoid curve hits.

However when the plateau hits, there will be avenue for another sigmoid curve to build up on

>>57399017
>Now that Moore's Law is super fucking dead,
Moore's Law never, ever existed except in the monds of some tech-press drones and the people who lap up their rubbish.
Gordon Moore made and observation; it was never a law as there was no way it could be enforced by either man or nature.
He observed that the density of transistors on a chip seemed to double periodically. This may have been true for a short time during the early days of Intel but outside of that it was never true. Lazy journalists made it fit.

>>57399137
>he thinks free means competitive.

>>57404209
W-what? You can't be serious? What's next, you're gonna tell us that Godwin's Law is not really a law?

>>57404252
Looks like Poe's Law isn't actually a law either. Who'da thunk?

>>57399766
>Moore's Law
>HDD

>>57399017
>actual tech thread
>abandoned after 24 replies

>/g/ - Technology

File: synapse-chip-infographic_hires_nobuttons.jpg (4MB, 2538x8725px) Image search: [Google]

4MB, 2538x8725px

>>57399017
>why not 3D print genetically modified brains to specifications
But IBM is doing basically just that

>>57408164
>putting a neural network on a CPU

The future of machine learning is looking gud

>>57407934
/g/ - battlestations and programming homework

>>57408310
I hope our new robotic overlords like me.