The Storm Approaches

Seriously guys, you need to wake up. Computing is going to change very rapidly in the next couple of years. I'm telling you this because I care about you guys. The other plebs, web monkeys, and enterprise jocks don't that this is going to make them obsolete yet. This is your chance to get ahead.

https://www.youtube.com/watch?v=JzbMSR9vA-c

Most people commenting on that video don't understand the true ramifications and horrors it will bring to them. The Machine will make it possible for an adept /prog/rider to master the world around them and exit from the shackles of the State. We're talking about recursively self-improving, super-intelligence levels of fire power here.

Memristors are at the doorstep, and with that will come machines that can solve NP-hard problems in P time. A computer architecture that doesn't need to fetch data from secondary memory into core memory, and write it back out. No more N Turing read/write heads accessing M memory cells, where N < M. Rather it will be N = M. This is a realm where no longer is there a distinction between a processor and a memory cell, but where a cell can polymorphically function as either in addition to programming the behaviour of neighbouring connected cells.

We're about to unlock the true magical power of computation. The true power underlying all of reality. This shit with classical Turing machines will be looked back upon with a nostalgic humor and disdain by the Gods of the future.

"Those plebs really enjoyed wasting their lives programming Touring machines! Can you believe it!"

Gather your courage and head into the storm before it reaches the common peasant. You shall become as Gods. The stragglers and slow-adoptors will perish. Non-programmers will become your thralls.

>>1

Too much hype. Yes it allows solving 2^N problem in N steps. But it still requires 2^N memory. And the technology is still in early stages. Decades will pass, before it gets adopted by the industry and NVidia releases a GPU supporting hysteresis computing elements.

On the other hand, Quantum Computing reduces the memory usage too!

plebs

Go back to 4chan. Computation is not underlying reality and P ≠ NP. An infinite amount of computational power will not overcome this or be able to do things that only human minds can. Hard AI isn't possible either.

If they can't even write the ``machine os'' in lisp then it isn't anything special.

>>2
Did they say 2^n memory? The UMM stuff has only polynomial time and space needs for NP.

>>5

The idea is that memory does computation itself, without an external CPU. Yet external CPU is still required to setup what computation will be done by each memory cell - meaning you still need to visit every cell.

>>6
That sounds like space = time to me. If so, p time implies p space.

>>2

But it still requires 2^N memory

No, it can get do this with information overhead.

Quantum Computing reduces the memory usage too!

It's more powerful than a mere probabilistic Turing machine.

>>3

and P ≠ NP

That only holds on a Turing Machine, not a non-deterministic Turing machine. A non-deterministic Turing machine is capable of solving NP problems in P time with P number of resources.

>>5
Exactly.

>>6
You still need a central unit to coordinate the clock impulses, but that's it. The memory cells do the computation and communicate with their connected neighbours.

>>8

>>3

and P ≠ NP

That only holds on a Turing Machine, not a non-deterministic Turing machine. A non-deterministic Turing machine is capable of solving NP problems in P time with P number of resources.

Not quite. It's stronger than that. NP means nondeterministic polynomial, not nonpolynomial (which people use as a shorthand or out of ignorance.) So computing NP in P on an NDTM doesn't change anything; NP is meant to express what an NDTM can compute in P.

You think you can write the software of the future that will be of importance? You think you can change the future with your programs? You think creating your little projects on Github, written in your toy languages is going to amount to anything?

No, the only way you can program something is to be a /prog/rider and prepare for non-deterministic Turing computation. So study memristors, study memcomputing.

What do you do? You are pacified. You eat your cheese-flavored pasta. You program your Touring-complete garbage collected programming languages. You say we should masturbate to Touhou. You say we should go to the LISP meetup groups. You are not in touch with reality. I have no interest in what you have to say.

You are cattle. You are zombies. You are single celled organisms.

Admit you are a slave.

Because you want to be a slave. Because you are a masochist. You are pathetic. You are weak. You are not going to write the software of the future. You are not going to do anything. Because you are fed an opiate. You are under an illusion.

If you're a /prog/rider, you will stand up and accept reality. If you really want to write software that will matter, you will man-up and learn how to program Universal Memcomputing Machines.

>>3

P ≠ NP

PROVE IT YOU STUPID NIGGER MOTHERFUCKER

>>8

You still need a central unit to coordinate the clock impulses, but that's it. The memory cells do the computation and communicate with their connected neighbours.

You also need CPU to load the data from secondary storage. Then again, most computer games for example have huge world, which cannot be stored in RAM. Moreover, tasks beyond simple search and sorting require coordination and long "X depends on Y" chains on stack, which cannot be easily implementing through hysteresis (how would you implement CONS-lists using memcomputing?). And so you got your bottleneck again.

Also, the term "hysteresis" reminds me of "histomorphism" from Haskell - another over-hyped snake-oil solution to parallel computation (glasgow crooks promised that immutability would give implicit parallelism)

Implementation level specifics will change only the lowest level of the programming model. You can easily write a C compiler that generates memcomputing instructions.

>>12

You also need CPU to load the data from secondary storage.

There is no secondary storage. That's why it's such a big deal. You have petabytes of compute-memory.

"hysteresis"

IHBT. You might as well complain about source code character set impacting the performance of the compiled code.

>>14

Nonvolatile information storage requires the existence of energy barriers that separate distinct memory states from each other.[4][5] Memristors whose resistance (memory) states depend only on the current (like the HP memristor) or voltage history would be unable to protect their memory states against unavoidable fluctuations and thus permanently suffer information loss: the proposed hypothetical concept provides no physical mechanism enabling such systems to retain memory states after the applied current or voltage stress is removed. Such elements can therefore not exist, as they would always be susceptible to a so-called "stochastic catastrophe".[5]

>>15
I don't buy it, what are your references?

>>1

Memristors are at the doorstep, and with that will come machines that can solve NP-hard problems in P time.

What the fuck are you smoking? Go back to Reddit, Kurzweil. I'm sick of your Reddit hype.

https://www.youtube.com/watch?v=JzbMSR9vA-c

I'm now at about 20 minutes in and it looks like a fucking Apple keynote. People are literally clapping over colorful plastic pieces. They have nothing to show but a few pictures and over-simplified diagrams.
Also, they didn't invent RISC (and definitely not in the mid-80s).

>>18

Yeah. Every self-respecting computer user should know that the first RISC design (MIPS) was implemented by John L. Hennessy, it required him to solve the problem with interlocked pipeline stages, which are reflected in branch delay slot (i.e. you get a free slot, while CPU fetches the shit).

/polecat kebabs/

>>19
I remember reading that MIPS has better processors and instructions but people still use ARM because they are hipsters

>>21

MIPS was pretty efficient, so that all successful video game console and workstations used it back then. Instruction set also had this little cuteness feel to it, where all was seemingly logical and solidly designed.

>>3

Hard AI isn't possible either.

Theist pseudointelligent cocksucker detected.

>>23
Atheist pseudointelligent feminist submissive python/ruby/rust/javascript/java user faggot subhuman non-aryan black nigger jewish-foreskin-sucker detected.

>>10

You say we should go to the LISP meetup groups. […] I have no interest in what you have to say.

That's a shame because The Machine looks like a really cool platform for a Lisp OS with its heavily simplified memory hierarchy.

However, having non-volatile memory on oxygen ion basis sounds paradoxical to my pseudo-scientific mind.

>>25

First you provide us with "The Machine" specification and emulator.

>>10

Touring-complete...

keep misreading this as Touhou-complete

>>27
>le pedophile dyslexic

This thread reminds me of 4chan, go back there.

>>29
To have something reminding you 4chan that means that you are from there
Get back to 4chan and stop spamming /prog/ with your shit

>>28-30
Shut the fuck up already, nobody cares.

>>29-31
Shut the fuck up already, nobody cares.

The Pleasure Of Being Cummed Inside.

How do you pronounce TPOBCI?

1) Tu Pob Ki
2) Tupabki
3) To bachi (faggot)
4) other

>>34
2PAC

4)
too-pob-ski

Memristors are at the doorstep, and with that will come machines that can solve NP-hard problems in P time.
machines that can solve NP-hard problems in P time.

Never going to happen. I don't care how powerful you think Memristors are going to be. Not even quantum computers are able to solve NP-hard problems in P time.

>>37

human brain can prove math theorem, which are NP-hard

>>38
Human brains also do not calculate billions of addition/substraction/etc equations every second like a logic circuit does. The human brain is a pattern matcher that's been forged through evoluctionary pressures and not designed as brute number crunchers.

>>38
The human brain operates like a stochastic pathfinder, not a closed-form solver. Giving an NP-hard problem to a human brain does not guarantee you a solution in P time, not even if you're some kind of math genius.