CES: Nvidia launches Tegra K1, based on Kepler

[Attreyu]

This means we'll begin to see powerful games, at PC level, running directly on mobile devices. They demoed Battlefield 3 and Unreal Engine 4 on this new architecture. So from this year on, apart from the PC and the consoles, AAA games will be released for mobile devices as well.

I wonder how will this impact VR gaming. IMHO it opens the gates for an influx of adapters like Vrase, Durovis Dive and 3Deva while making it a must for Oculus to release a completely untethered mobile Rift, centered around Android. And even then, I wonder how it would compete with the continous advances in hardware cappabilites (2K displays soon to be followed by 4K ones, excellent sensors, etc.) from the mobile industry and keeping their edge.

The competition just got harder. We're going full mobile.

http://www.engadget.com/2014/01/05/nvidia-tegra-5/

"In what's becoming a yearly tradition, NVIDIA came to its CES press conference with tidings of its next-generation mobile processor which will begin gracing devices in 2014. The chipset maker officially announced Tegra K1, which features the first 192-core GPU based on the Kepler architecture which was originally designed for desktops and notebooks and later added to supercomputers. As you can imagine, NVIDIA is bringing its graphics chops to the new DirectX 11-powered GPU, and it packs a serious punch -- so much so, in fact, that it will come with support for Unreal Engine 4 and OpenGL 4.4. In the company's usual form, we were treated to demos of the new chip's power compared side-by-side with a Tegra 4, and the difference was quite noticeable; the K1 offers real-time computing, global illumination, higher dynamic range and greater detail like reflective surfaces, dripping water and other realistic physically-based rendering. Interestingly, Tegra K1 will actually come in two different versions: a 32-bit option with a 2.3GHz "4-plus-1" A15 CPU and a 64-bit unit with a 2.5GHz dual-core Denver CPU. Naturally we're a bit more curious about the latter at the moment, and we'll update you as we get more information about either chip -- such as availability.

Update: We received a little extra clarification from NVIDIA reps, who have told us that both versions of the K1 have already been certified by AT&T and Vodafone (among other carriers), and we should expect to see devices with the new chipset near the end of the first half of 2014. We also learned that although LTE support doesn't come natively on the chipset, it'll still be available thanks to an external chip that will be part of the K1 setup. It's hard to say whether or not this will help NVIDIA gain some lost ground on Qualcomm, but only time will tell."

[Popopinsel]

Sounds like a beast on paper but from these rather simple and static environments it seems to perform pretty bad (for our VR purposes) in its current state. Sub-30 fps VR experience?

But perhaps it's just due to a crappy Fraps screen captuure...

[MSat]

Sounds like a beast on paper but from these rather simple and static environments it seems to perform pretty bad (for our VR purposes) in its current state. Sub-30 fps VR experience?

But perhaps it's just due to a crappy Fraps screen captuure...

It might be fine as long as you're not expecting cutting-edge graphics.

[Likay]

Do anyone know a list of devices which will have tegra? Just so i will not buy one by mistake.

[Attreyu]

It's a Kepler. If the CPUs keep up with the chipset, devices using it should have no problem running today's games at 60FPS or more.
Keep in mind that it's going to be in 64bit too - this means more RAM. Samsung F series will be unveiled next month at WMC - 8GB RAM.
They're closing the gap, and fast. This year's going to be awesome for mobile VR. And there's no other natural way for VR but mobile and untethered.

[MSat]

Sure, it's Kepler, but it's intended for mobile - meaning that in order to keep power usage down you have to either keep the clock down, keep the number of shader units down, or both. Otherwise you end up with a laptop or desktop-class GPU with integrated CPUs. If that's what you want, may I recommend an AMD APU? There's no free lunch to be had other than what might be attained from new CPU/GPU designs and smaller technology nodes (though not necessarily true).

[Attreyu]

You're absolutely right, of course. But even with a lower shader or clock speed, it's still miles better than what we have at the moment.
I mean, I have a laptop with Optimus - it's a 640m and it runs BF3 at 1080p like butter, not to mention Source games.
One year from now they should have no problem running anything pre 2013 at 2K resolution with everything turned up. After all, even the PS3 graphics which are what, 5 years old ? can still pack a punch graphics wise.

[colocolo]

Battlefield 3 on SoC Rift solution would be great.
We shouldn't forget that there is also a world that just can't afford next gen games or even a PC.
Those ones (China , South America) could really profit of an all-in-one solution.
Lets spend those 20 bugs extra so that at least a major part of the world has the chance to get their hands onto that marvellous VR.

[Attreyu]

Battlefield 3 on SoC Rift solution would be great.
We shouldn't forget that there is also a world that just can't afford next gen games or even a PC.
Those ones (China , South America) could really profit of an all-in-one solution.
Lets spend those 20 bugs extra so that at least a major part of the world has the chance to get their hands onto that marvellous VR.

Absolutely. The downside is that you still need to pay something for the tablet/smartphone. Of course, not as much as you would on a PC+your chosen VR solution.

[colocolo]

Battlefield 3 on SoC Rift solution would be great.
We shouldn't forget that there is also a world that just can't afford next gen games or even a PC.
Those ones (China , South America) could really profit of an all-in-one solution.
Lets spend those 20 bugs extra so that at least a major part of the world has the chance to get their hands onto that marvellous VR.

Absolutely. The downside is that you still need to pay something for the tablet/smartphone. Of course, not as much as you would on a PC+your chosen VR solution.

I actually meant the consumer Rift to encorporate a Tegra K1 + VR home base level(UE4).
(for example you pay right now 25$ for Tegra 4 if you take 1000)

https://www.youtube.com/watch?v=atVxxy0pwKA

[Attreyu]

I actually meant the consumer Rift to encorporate a Tegra K1 + VR home base level(UE4).
(for example you pay right now 25$ for Tegra 4 if you take 1000)

You would end up paying 400$+ for something that can't do everything a tablet with Tegra4 can and costs about half the price. It would be like buying a Shield when you already have a Nexus 7 and a bluetooth controller

My point is we're better off using what we already have and breathe new life into it. I'm not going to spend hundreds upon hundreds of dollars on something extra when I can buy a tablet or smartphone with a Tegra K1 or a Snapdragon 805 and have everything that a PC can offer for a fraction of the price.

[TheHolyChicken]

My point is we're better off using what we already have and breathe new life into it. I'm not going to spend hundreds upon hundreds of dollars on something extra when I can buy a tablet or smartphone with a Tegra K1 or a Snapdragon 805 and have everything that a PC can offer for a fraction of the price.

I don't understand why you're convinced tablets/phones, plus a set of lenses, is the right way to go. Tablets/phones have the issue of:

• Poor/no positional tracking
• Slow/inaccurate head orientation tracking (phone gyroscopes may be getting better, yes, but there's no incentive for a phone to have as good a gyroscope as we want in an HMD)
• A panel that will be inferior to a zero-persistence panel that looks likely to be in the Rift, meaning we have smeary visuals when headtracking
• A panel that may look poor under magnification
• A panel that is likely to have a lower refresh rate than the Rift
• Heavier than a standalone HMD
• The weight distribution will be poor, as the heaviest part (the phone/tablet) will be furthest from your head

It's just going to be a bad experience. The only, and I repeat, the ONLY benefit of using a tablet/phone, plus third party lenses, is that it'd be cheaper for someone that already has a suitable tablet/phone (and probably more expensive for someone that does not). Apple prove time and time again that people are happy to pay for a well-packaged and slick product that gives a good user experience, even if it costs more than the alternatives. Phone/Tablet + lenses is a cheaper, but crappy experience. I'm just not seeing it being successful, and a far cry from being the preferred option.

[colocolo]

I actually meant the consumer Rift to encorporate a Tegra K1 + VR home base level(UE4).
(for example you pay right now 25$ for Tegra 4 if you take 1000)

You would end up paying 400$+ for something that can't do everything a tablet with Tegra4 can and costs about half the price. It would be like buying a Shield when you already have a Nexus 7 and a bluetooth controller

My point is we're better off using what we already have and breathe new life into it. I'm not going to spend hundreds upon hundreds of dollars on something extra when I can buy a tablet or smartphone with a Tegra K1 or a Snapdragon 805 and have everything that a PC can offer for a fraction of the price.

like HolyChicken in detail explained. A phablet VR experience wont be the same as a superior Rift experience.
Those adapters will have their niche though for people who just want to try it out.
People who seriously want to get into VR stuff are going to buy the Rift because they just don't want to get sick of it.
Big smartphone manufacturers are far from releasing a VR version of their smartphones as SONY impressively proved yesterday.
Oculus VR has no competitors(Valve is more a healthy coexistence to drive the PC-VR revolution), plain and simple. They will rule this field not only because they are eager to offer a version for perhaps 250 $!
By the way i don't even possess a smartphone and won't spend my money on one in the future but rather for a Oculus Rift V1/V2, a new high end graphic cards or a neural interface.
To the devil with those smartphones, people on the street look like zombies.

[Attreyu]

My point is we're better off using what we already have and breathe new life into it. I'm not going to spend hundreds upon hundreds of dollars on something extra when I can buy a tablet or smartphone with a Tegra K1 or a Snapdragon 805 and have everything that a PC can offer for a fraction of the price.

I don't understand why you're convinced tablets/phones, plus a set of lenses, is the right way to go.

I'm considering them cheaper alternatives. I'm convinced of it because I've tried them both and there's no difference. In fact, tablets are better.

Tablets/phones have the issue of:

• Poor/no positional tracking

Rift has no positional tracking. There are almost no games with positional tracking implemented. Aside from weapon movement, there isn't any reason to implement positional tracking at all. Even if you need it, it's the ODT's job to do positional tracking, not the HMDs. We also have Kinect, Hydra and a dozen more device which can do that.

[*]Slow/inaccurate head orientation tracking (phone gyroscopes may be getting better, yes, but there's no incentive for a phone to have as good a gyroscope as we want in an HMD)

It's not slow, nor innacurate. Have you ever used FreePie or a VR app with integrated headtracking like the ones from Durovis ? I haven't sensed the difference, the headtracking is spot on.

[*]A panel that will be inferior to a zero-persistence panel that looks likely to be in the Rift, meaning we have smeary visuals when headtracking

I don't know what will be in the Rift in 2015, but we already have awesome displays of all shapes, types and sizes NOW. They're 1080p and they're gorgeous when playing.

[*]A panel that may look poor under magnification

See above. More, the Rift has 720p resolution. Nothing to compare here, this year's tablet displays destroy it.

[*]A panel that is likely to have a lower refresh rate than the Rift

Again, see above.

[*]Heavier than a standalone HMD

An adapter + phone weighs about 200g. An adapter + tablet weighs about 350g. A Rift weighs more than 400g.

[*]The weight distribution will be poor, as the heaviest part (the phone/tablet) will be furthest from your head[/list]

And that's different from the Rift how ? The heaviest part is the display, and it's 5-7 cm away from you. If you use a smartphone you can't even compare the two.

It's just going to be a bad experience. The only, and I repeat, the ONLY benefit of using a tablet/phone, plus third party lenses, is that it'd be cheaper for someone that already has a suitable tablet/phone (and probably more expensive for someone that does not). Apple prove time and time again that people are happy to pay for a well-packaged and slick product that gives a good user experience, even if it costs more than the alternatives. Phone/Tablet + lenses is a cheaper, but crappy experience. I'm just not seeing it being successful, and a far cry from being the preferred option.

I'm not going to enter into an Apple dispute but I don't know even 5 people who would buy a 300$ niche gadget to play a VR game. The Rift is out since almost two years and they have a tiny market pool which is not going to get radically larger.
But I know hundreds of people who would spend 70-100$ on something that can breathe new life and scope into their smartphones or tablets without breaking a sweat.

VR is not for geeks, its supposed to reach mainstream. Games, movies, AR for everybody.

[Attreyu]

like HolyChicken in detail explained. A phablet VR experience wont be the same as a superior Rift experience.
Those adapters will have their niche though for people who just want to try it out.
People who seriously want to get into VR stuff are going to buy the Rift because they just don't want to get sick of it.

I see the Rift as a niche and a phablet as mainstream.

And I don't believe VR is for the geeks and "serious" people. It's for everybody

If you care about technicalities, sure, buy a Rift. If you don't, buy a Vrase or something like that.

Not everyone cares to drive a Porsche or Mercedes, or generally pay a premium for something they can get for a fraction of the price.

If you can offer a wireless, headache-free experience to someone, it's a step forward.

[colocolo]

i didn't said that VR is only for the geeks or the serious.
People who want to get a taste of it will invest in adapter first and afterwards if they get really interested because of this experience they will consider 'seriously' buying a Rift because of the benefits they'll get from it. (Nausea=no fun primarily)
good VR is not a luxury but rather sth like a good healthcare.
Where is the comparision between a Ferrari and a Seat? Both are superior products. Only difference the Ferrari is excessively expensive sold as a luxury product. Same with Audi,BMW, Mercedes. (oh yeah german engineer ingenuity ...phhh...)
The point is that the Rift is really not a luxury product. Oculus VR is doing everything to sell it as cheap as possible.
And by the way even if we get those smartphones for 0 dollar it doesn't mean that they are a present.
25-35€ per month still is a lot of money in 24 months(600-840). You can easily make your phone calls with your old phone and only spend like 10 bugs for them per month.(240) The difference makes a Rift.

[Attreyu]

i didn't said that VR is only for the geeks or the serious.
People who want to get a taste of it will invest in adapter first and afterwards if they get really interested because of this experience they will consider 'seriously' buying a Rift because of the benefits they'll get from it. (Nausea=no fun primarily)
good VR is not a luxury but rather sth like a good healthcare.
Where is the comparision between a Ferrari and a Seat? Both are superior products. Only difference the Ferrari is excessively expensive sold as a luxury product. Same with Audi,BMW, Mercedes. (oh yeah german engineer ingenuity ...phhh...)

I agree with everything you've said.

The point is that the Rift is really not a luxury product. Oculus VR is doing everything to sell it as cheap as possible.

I haven't said it's a luxury, I've said it's a premium product for a niche segment. I'm sure they are trying to sell it as cheap as possible, without cutting any corners. You have the best specs in a HMD, and best minds working on it. That's beyond doubt.

And by the way even if we get those smartphones for 0 dollar it doesn't mean that they are a present.
25-35€ per month still is a lot of money in 24 months(600-840). You can easily make your phone calls with your old phone and only spend like 10 bugs for them per month.(240) The difference makes a Rift.

No, but you have them already. Adding VR is just the icing on the cake. You can't talk about the difference because you need a phone. You haven't bought a nice phone because you had VR in mind, but because you use it for other purposes already.

If you want to buy a nice phone and a nice HMD, you would end up spending 1000$+. Serious and passionate peole would do it. The rest of us, not so much, especially when you are presented with the alternatives.

[TheHolyChicken]

...

I'm not comparing the phablet + adapter experience against the Rift dev kit - I'm comparing against the expected specs/experience of the consumer Rift. I fully accept that a phablet + adapter solution is probably better overall than the Rift is right now, but that won't last for long; the consumer Rift is going to have specialization that a phone cannot match (positional tracking, low-persistence, high refresh rate), giving a greater experience than a phone will be capable of.

There are almost no games with positional tracking implemented.

YET. Of course there's hardly any software out there with positional tracking - the Rift does not yet have it! I fully expect that Rift titles will be promptly updated with positional tracking once the SDK is updated.

Aside from weapon movement, there isn't any reason to implement positional tracking at all.

Positional tracking is essential to track head properly, so that there are not large conflicts between estimated eye position and actual eye position - a huge culprit of nausea. Weapons have nothing to do with it. I've shown the Rift to about 50 people at this point, and while virtually all of them have been impressed and excited, a large number have also unfortunately come away a little woozy. Good positional tracking is the key to alleviating this issue.

Even if you need it, it's the ODT's job to do positional tracking, not the HMDs. We also have Kinect, Hydra and a dozen more device which can do that.

Relying on third-party peripherals/tech is bad. Increases costs for the consumer having to buy multiple products, adds complication for the consumer, and segregates the target market. You just said yourself that there's almost no games with positional tracking implemented; this is the reason. Why aren't many developers already using the Kinect, Hydra, or one of the "dozen more devices" that could do this?

[*]Heavier than a standalone HMD

An adapter + phone weighs about 200g. An adapter + tablet weighs about 350g. A Rift weighs more than 400g.

This is probably a good point to mention that there's hardly any phones/phablets out there that even meet the desired size we would like for our optics to perform best, which is ~6-7". The largest that the VrAse accepts is the Galaxy Note II (weighing in at 183g), and that only has a 5.5" panel. An adapter solution is either going to be heavier, or be optically inferior.

[*]The weight distribution will be poor, as the heaviest part (the phone/tablet) will be furthest from your head[/list]

And that's different from the Rift how ? The heaviest part is the display, and it's 5-7 cm away from you. If you use a smartphone you can't even compare the two.

An entire phone/phablet weighs less than just its panel.

But I know hundreds of people who would spend 70-100$ on something that can breathe new life and scope into their smartphones or tablets without breaking a sweat.

I'm sure plenty of people would buy a VrAse-type device to watch films on. If I did a lot of travelling I would probably go and buy one right now. I can't see it being at all comparable for VR, though.

[colocolo]

But nevertheless the performance of the new K1 is impressive.
I remember playing Crysis 2 with a hd 5450 3 years ago at 11 FPS. HAHA. it was fun! see what immersion can do?
The 5450 contains 292 million transistors , 80 unified shader units.
It would be a hoax to see this tiny efficient processor executing Crysis 2 at a reasonable frame rate.
Trine 2 they showed at the conference, a beautiful puzzle game based on UE3(demo downloadable at steam) , isn't that unassuming as it seems. My amd quad core hd 6850 powered PC reaches 100 FPS but consumes like 200 watts.
Its pretty impressing what this tiny chip does.

[Attreyu]

...

I'm not comparing the phablet + adapter experience against the Rift dev kit - I'm comparing against the expected specs/experience of the consumer Rift. I fully accept that a phablet + adapter solution is probably better overall than the Rift is right now, but that won't last for long; the consumer Rift is going to have specialization that a phone cannot match (positional tracking, low-persistence, high refresh rate), giving a greater experience than a phone will be capable of.

As now almost every smartphone/tablet surpasses the Rift, by the time they'll launch, the gap will be even bigger. They're better now, they'll be better still in the future. No sole device can compete with the entire industry in terms of hardware and software evolution.

There are almost no games with positional tracking implemented.

YET. Of course there's hardly any software out there with positional tracking - the Rift does not yet have it! I fully expect that Rift titles will be promptly updated with positional tracking once the SDK is updated.

So what are we talking about ? Are you comparing some yet not launched product + your own wishes with already launched products that are already better ? Even so, positional tracking is something that can be implemented in software as long as you have the necessary sensors. And we have. And we will.

Aside from weapon movement, there isn't any reason to implement positional tracking at all.

Positional tracking is essential to track head properly, so that there are not large conflicts between estimated eye position and actual eye position - a huge culprit of nausea. Weapons have nothing to do with it. I've shown the Rift to about 50 people at this point, and while virtually all of them have been impressed and excited, a large number have also unfortunately come away a little woozy. Good positional tracking is the key to alleviating this issue.

Ok, point taken. Positional tracking will be implemented. It's not something that it can't be done, so it's not something inherently achievable only by the Rift. Like every other characterstic of the Rift, with the added benefit of eliminating the cables. And the price.

Even if you need it, it's the ODT's job to do positional tracking, not the HMDs. We also have Kinect, Hydra and a dozen more device which can do that.

Relying on third-party peripherals/tech is bad. Increases costs for the consumer having to buy multiple products, adds complication for the consumer, and segregates the target market. You just said yourself that there's almost no games with positional tracking implemented; this is the reason. Why aren't many developers already using the Kinect, Hydra, or one of the "dozen more devices" that could do this?

See the above. Soon the mobile HMD adapters will probably have this cappability, if it results to be absolutely needed.

[*]Heavier than a standalone HMD - An adapter + phone weighs about 200g. An adapter + tablet weighs about 350g. A Rift weighs more than 400g. This is probably a good point to mention that there's hardly any phones/phablets out there that even meet the desired size we would like for our optics to perform best, which is ~6-7". The largest that the VrAse accepts is the Galaxy Note II (weighing in at 183g), and that only has a 5.5" panel. An adapter solution is either going to be heavier, or be optically inferior.

We have displays across the range, from 5" to 7" and higher. My own adapter supports them all up to 7". More than that and you'd start having weight problems and the display is too big, the human eye can only cover so much.

[*]The weight distribution will be poor, as the heaviest part (the phone/tablet) will be furthest from your head[/list]
And that's different from the Rift how ? The heaviest part is the display, and it's 5-7 cm away from you. If you use a smartphone you can't even compare the two.

An entire phone/phablet weighs less than just its panel.

Indeed, so it's more lightweight than a Rift, no matter how you consider it. Rest assured that it's a perfectly comfortable experience, much more than the Rift and its 400 grams + cables dangling around your neck.

But I know hundreds of people who would spend 70-100$ on something that can breathe new life and scope into their smartphones or tablets without breaking a sweat.

I'm sure plenty of people would buy a VrAse-type device to watch films on. If I did a lot of travelling I would probably go and buy one right now. I can't see it being at all comparable for VR, though.

What you can't see doesn't mean it's true. I can see it perfectly, comparing them one spec at a time on paper and in real life.

The only thing still in Rift's favor, and that's about to change (like everything else) is the latency. But we're almost there, devices like Nvidia Shield are proving it now. By next year this would have been a non issue.

Comparing the Rift and the mobile adapter alternatives is really like comparing a PC with a tablet. In the end, the tablet's going to win in all fronts. We're half way there and the gap is closing really fast.

One year from now, you can take your still officially unreleased 300$ Rift + 1000$ PC and its positional tracking, cables, superior optics and FOV - you'll be able to play, tethered to a big, powerfull PV and pretty much that's it.
And I'll take my 70$ adapter+230$ tablet with 4K display and 2" lenses, completely untethered, go outside on the porch, play the same games you play and do a ton of other things on the side. While dancing around your overpriced setup

[thatdude90210]

Beyond the fact that a PC will produce better graphics, I'm not sure I want a mobile VR device. Who wants to be essentially blindfolded in public. You are just asking to have something stolen, or worst.

[colocolo]

Beyond the fact that a PC will produce better graphics, I'm not sure I want a mobile VR device. Who wants to be essentially blindfolded in public. You are just asking to have something stolen, or worst.

Its to be understood as a nice addition, nothing more. Those chips cost like 20$. Its worth it.

[Attreyu]

Beyond the fact that a PC will produce better graphics, I'm not sure I want a mobile VR device. Who wants to be essentially blindfolded in public. You are just asking to have something stolen, or worst.

Why use them in public, they're made for home/office use, for when you want to play from the couch, another room or simply to watch a movie in bed.

[colocolo]

so what i understand from the sheets they presented is that a K1 has possibly double the speed of a PS3 or Xbox360.
Rage ran on aPS3 at 60FPS at 720p.
Maybe we will see some serious VR games(like rage) tetherless if Oculus is not already making deals with Nvidia.
Nvidia will have a hard time with AMD new API and console chips, they need customers.

[GeraldT]

Has anyone seen the Glyph demonstration with (iirc) an iPhone?

That is how I imagine the most popular Rift scenario will be in 2-3 years ... you plug in your smartphone, so you don't have to buy a new Rift just because the processor is outdated. You reach all those consumers that switched from PC to more modern platforms like a tablet or powerful smartphone. You give a perfect alternative to occupying the tv (so your wife/girlfriend ...) can watch wannabe models/superstars and still get a great gaming sensation (you can even sit in the same room if you dare).

Seeing how powerful mobile devices are getting is great. That said ... we won't play Crysis 4 VR on a smartphone the same way we play it on PC. For the progamer looking for the best possible experience the PC will stay the platform of choice for a long time I am sure. You just can't put enough power into those small devices yet and several examples optimized to look great won't change that.

It's the age of VR and with that it will likely be a great time for indies looking for ways to create awesome experience on limited platforms like mobile while AAA can build the most awesome VR Battlefields on PC