NVIDIA's Foveated Rendering

[Kim Jong-Un's Pajama Pants]

• Tweet

This rendering uses where you are looking and renders the sharpest image in that location.  Like in real life.  This saves processing power, which would otherwise be used to sharply render images that you are not even looking at. This can allow for more detail in virtual reality games.

It has shortcomings, that include noticeable distractions.  NVIDIA has some interesting solutions.

https://www.roadtovr.com/nvidia-perceptually-based-foveated-rendering-research/

[Nefertiti]

Interesting - NVIDIA is impressive. I have been wanting to check out Shield TV's 3D effects - if it works without the headsets & stuff. If they integrate GPUs into daily devices - phones, etc - without the driver downloads & too much bother - perhaps autopush via OTA. They are a visual computing house afterall - they need to break out of the gaming mindset - it is a narrow space.

I wonder what they are upto with AI? - besides autonomous vehicles. They lost the autonomous camera race.

[RV Pundit]

Interesting.

[Nefertiti]

These "smart" devices - AI, VR, autonomous, etc - are the "internet of things". They do such heavy computing that they cannot rely on the data center or cloud,  because that kind of latency is impossible in conventional technology. That amount of secure bandwidth does not exist- not even 5G can hack it. 

It is what I was discussing with MOON Ki & Windy on some thread - quantum computing - heavy computing on consumer devices can only be accomplished by quantum. And MOON Ki mocked me on the quantum chip - and dismissed quantum cryptography.

Computing, connectivity, energy, etc - vast sectors are stuck for decades due to tech challenges - that quantum can unlock. Inviability of wireless power is principally a bandwith issue - because the electromagnetic spectrum is not yet properly "hacked" or understood - so the gamma rays can be used as super-voltage carrier. Instead at present we only use higher frequencies to transmit radio, tv & such low-end signals. Once we figure out how to transmit electricity - using gamma - without striking every living cell dead or cancerous - gameshot.

Well, quantum can do that and more. But first of course it has to be used in edge computing - which is IoT - where we have driverless cars, smart homes, robots, and all - before going bandwidth because all these mobile data centers have to connect faster. And securely. The natural progress will be to transmit power - and solar megapanels will be hoisted pronto on satellite to tap unfiltered solar.

This is the only way of course Mars becomes viable - cause the frozen atmosphere can be inhabited with boundless power sources. veri need not worry so much or claim Elon Musk is crazy.

[MOON Ki]

It is what I was discussing with MOON Ki & Windy on some thread - quantum computing - heavy computing on consumer devices can only be accomplished by quantum. And MOON Ki mocked me on the quantum chip - and dismissed quantum cryptography.

I did not mock you; I simply disagreed with you.  Nor did I dismiss quantum cryptography.      Generally, there were two points where we disagreed:

First, what the Chinese had demonstrated was simply a link that uses quantum technology, and you seemed to jump from what is simply an interconnection to quantum computing and its power.    Second, you attributed an extraordinary role to quantum computing in cryptography.   

You wrote:

Honestly current computing is stretched - cryptosystems & such heavy stuff - it's why WhatsApp, Messenger or GMail messages are "encrypted" on the server or cloud or whatever .... If quantum was widely available simple gadgets would be able to handle it.   Offer people a real hack-proof consumer device and and there's your big reap.

First, the point that I emphasized was that encryption does not actually require that much computing power; quantum computing---not the links (Chinese style)---is relevant in encryption only in that quantum computers will someday have sufficient power to break most current cryptosystems.  The relevance of the Chinese style "quantum links" are that they will enable the detection of eavesdropping, not that they will help with encryption itself.  So their biggest use will probably be in transmitting encryption keys, not "content" itself, which in any case can be encrypted by other means (including future "quantum resistant" algorithms) so that eavesdropping doesn't do anyone any good.   Strictly, a great deal of what is passed under the label "quantum cryptography" actually has next-to-nothing to with encryption with quantum computers; much of it about the communication of encryption keys and the development of conventional "quantum-resistant" algorithms for conventional computers.   

Second, you mentioned things like Gmail as an example of where "current computing  ... is stretched ... server side".  To the extent that people want more security that is provided by the basic communication protocols used and wish to encrypt whatever they are sending from their laptop, cell phone, or whatever, they can already do so quite easily: Google supports that, as they explain here:

https://support.google.com/mail/answer/6330403?hl=en 

And, quite apart from all that, there is nothing---certainly not the lack of computing power---to stop a well-informed user from coding up his or her own cryptosystem, of just getting one on the web, agreeing on its use with the people he or she will be sending email to, and doing their own thing. Very little computing power required for that.   

Question: Just what is it that you think "quantum cryptography" will do and how?

Computing, connectivity, energy, etc - vast sectors are stuck for decades due to tech challenges - that quantum can unlock.  Well, quantum can do that and more.

Moore's Law---especially if one looks at it in terms of performance, not just numbers of transistors---has held for 50 years now.   So I'm not sure I would say computing is stuck.    Likewise, methods for interconnection have vastly improved.   And today we have low-power devices that might have seemed impossible not that long ago.  Of course, "stuck" may be considered a relative term ...

Well, quantum can do that and more. But first of course it has to be used in edge computing - which is IoT - where we have driverless cars, smart homes, robots, and all - before going bandwidth because all these mobile data centers have to connect faster. And securely. The natural progress will be to transmit power - and solar megapanels will be hoisted pronto on satellite to tap unfiltered solar.

I know little about the IoT.    But if someone were to ask me what I think will be a single prominent feature, my guess would be sensors---lots and lots of them.    I would also guess that, to go with those, the  biggest need will be for low-power and efficient distributed computing, rather than massive computing power (which will be required in relatively few places).

[Kim Jong-Un's Pajama Pants]

These "smart" devices - AI, VR, autonomous, etc - are the "internet of things". They do such heavy computing that they cannot rely on the data center or cloud,  because that kind of latency is impossible in conventional technology. That amount of secure bandwidth does not exist- not even 5G can hack it. 

I don't know if it is harder to secure data because it is more.  I don't think there is that correlation. 

Yes, latency and throughput can be relevant in apps.  If an app needs huge amounts of information, then it will suffer if that information is caught up on the network, regardless of how powerful the processor is.  The basic solution to that is just more bandwidth I think.

It is what I was discussing with MOON Ki & Windy on some thread - quantum computing - heavy computing on consumer devices can only be accomplished by quantum. And MOON Ki mocked me on the quantum chip - and dismissed quantum cryptography.

I notice he has responded to you.  But granted, quantum computing promises superior processing power.  This happens where actual processing is required.  I am thinking processor registers.

Computing, connectivity, energy, etc - vast sectors are stuck for decades due to tech challenges - that quantum can unlock. Inviability of wireless power is principally a bandwith issue - because the electromagnetic spectrum is not yet properly "hacked" or understood - so the gamma rays can be used as super-voltage carrier. Instead at present we only use higher frequencies to transmit radio, tv & such low-end signals. Once we figure out how to transmit electricity - using gamma - without striking every living cell dead or cancerous - gameshot.

That is above my paygrade.

Well, quantum can do that and more. But first of course it has to be used in edge computing - which is IoT - where we have driverless cars, smart homes, robots, and all - before going bandwidth because all these mobile data centers have to connect faster. And securely. The natural progress will be to transmit power - and solar megapanels will be hoisted pronto on satellite to tap unfiltered solar.

IoT is definitely the future and increasingly the present.  IoT, AI and Big Data can definitely benefit from quantum computing power.

This is the only way of course Mars becomes viable - cause the frozen atmosphere can be inhabited with boundless power sources. veri need not worry so much or claim Elon Musk is crazy.

Mars is an inhospitable place.  A hostile place.  As soon people land there, they will realize they are on a planet that is actively and vigorously trying to kill them in a million possible ways and with a very unfair chance of success.  One slip and you are dead.  It's more dangerous than being alone in a remote dersert or jungle or deep in the ocean or even the moon.  I enjoy reading the possibilities in Sci-fi.  But Sci-fi has a way of making stuff look easier.

[Kim Jong-Un's Pajama Pants]

Well, quantum can do that and more. But first of course it has to be used in edge computing - which is IoT - where we have driverless cars, smart homes, robots, and all - before going bandwidth because all these mobile data centers have to connect faster. And securely. The natural progress will be to transmit power - and solar megapanels will be hoisted pronto on satellite to tap unfiltered solar.

I know little about the IoT.    But if someone were to ask me what I think will be a single prominent feature, my guess would be sensors---lots and lots of them.    I would also guess that, to go with those, the  biggest need will be for low-power and efficient distributed computing, rather than massive computing power (which will be required in relatively few places).

Yes sensors.  And Big data.  And machine learning(ML).  Weather sensors, maybe millions of them, could gather information temperature, altitude, humidity, amount of dust etc in the atmosphere that needs to be processed by a gathering algorithm, normalized against historical data and provide near real-time feedback, for example.

[Nefertiti]

Nuh MOON Ki - you are both fundamentally wrong - Windy & you. Let me watch the speeches first.