It needs to eventually hit 8K or 16K per eye at high framerates, which is not impossible for mobile chips if foveated rendering and other advancements are used. Pack it all in a monolithic 3D chip for a big boost in processing power.
Sorry to burst your bubble, but the maximum resolution of today consumer GPUs is 7680 x 4320 at 60Hz.
Also, stacking dies for GPUs is not feasible unless you drop your power and frequency by at least half, which mean you would cut your performance at minimum by half of your silicon area. And if you stack 4+ dies together, you will drop the maximum performances per die even further due to heat which makes the whole concept totally non-viable financially. Basically, a double stack GPU would cost two time more and would barely provide 20-30% performance than a single die. Not to mention complexity in manufacturing such complex processors
"the maximum resolution of today consumer GPUs is 7680 x 4320 at 60Hz."
For a single output link. For multiple composite links (e.g. DP MST) that can be easily exceeded, and for internal rendering (i.e. without the PHY limitation) it is VERY easily exceeded, limited only by raw ROP capacity.
Based on the slow pace of product announcements and declining investment into R&D, it certainly looks like we are on the trailing end of our second VR fad. Maybe round three will finally bring us finely developed, useful products, but we still have not yet found broadly appealing use cases for teh technology yet so even if VR gets revived again in the future, it may still languish as a solution looking for a problem.
Welcome to being seven days late to the party. In any event, you are presenting pull side data, not push side product launches and investment which is what I was talking about a week ago. If you look at historic data for new product launches, you will see that pull peaks always trail push peaks, but the curves are generally mirrors of one another.
As far as I am aware, nobody has actually shipped a HMD based on the XR1 platform. Current non-devkit standalone platforms (e.g. Hololens, Oculus Quest and Go, HTC Vive Focus, Lenovo Mirage Solo) either use non-Qualcomm processors, or use use their own platform on a Snapdragon processor.
I have been searching and found the Vuzix M400, Glass Enterprise Edition 2 from google etc. They are not a consumer product, so I guess the XR1 platform was not exactly cheap to implement.
Maybe it's just me, but "XR" has the stench of death about it.
"VR" hardware is "good enough" for many purposes but the content is not there; it will never be the time suck that browsing the web, watching TV, reading, or playing conventional video games because you can't stand long sessions. The hardware is expensive, but if there was enough content it might seem look a good investment.
"AR" is not good enough because of the problems that Karl Guttag writes about on his blog: the optics are a tough problem.
"XR" seems to have been invented to fit in somewhere between "The Race to 5G" and "The Race to AI". I imagine that executives in boardroom are asking "What is our XR strategy?" Just as the chemistry department birthed a materials science center, and then a nanotechnology center, a new name can convince (some) people that something new is going on.
If a company can make a wide field of view AR display that can be switched into a VR mode by putting a cover on it to block light, they should try it. There is already effort underway to flatten VR headsets and retinal projection could accomplish both AR and VR in a compact form factor.
If it can be made indistinguishable from ordinary glasses or sunglasses (yes, including cameras), you can convince people to wear the AR device in public and switch it to VR mode when they are at home or somewhere they won't get mugged.
The "AR device that can be covered" will be a poorly performing VR device because the AR device has much more complex optics in the combiner.
If you want to augment your world here and now, the birdbath architecture outperforms the diffraction waveguide at low cost. The only drawback is that it is ugly.
To get people excited about the platform you have to be able to demo it, show it off on a sales floor, put it on somebody's head at the art exhibit. The game developers might have three of them for a team of ten such as artist, coder, tester and want to be able to show the game off easily to visitors.
For that, you have to have a headset that works with or without glasses. That's the minimum viable product. Once people are sold on the platform, then the market opens up for small/light/personalized.
There's no point in calling the death of VR for a second time yet. We haven't even had major game companies go all in on it yet. We've yet to see PSVR2, or a Nintendo device, for example.
At this point in the VR revival, that sort of thing really should have happened already in order to keep the industry healthy. It was in 2012 that Occulus first opened their Kickstarter which was, arguably, the beginning of enthusiam for a new generation of VR products. The lack of major effort by hardware manufacturers, console producers, and game developers across the board over the last half decad says quite a bit.
I don't think VR is "dead" but I also don't think the problem with VR is with the displays. Tracking could be better and more universal. Thus no need for a PSVR2.
Beat Saber is the Pong of VR. Somebody might put together a bestselling title if they can get the level of storytelling up to the level of Crazy Climber, never mind a Nep game. One of the most intriguing possibilities for VR is sharing a space with simulated characters, who can provoke strong feelings if they get too far from or too close to you.
If XR is the ecosystem of parts and systems software shared by AR and VR than it is a going concern. The silicon is mostly the same.
"XR" is the walking dead from a marketing perspective, however, because customers right now don't know what the value is in AR or VR and "XR" doesn't give any clarity.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
22 Comments
Back to Article
nandnandnand - Tuesday, February 18, 2020 - link
It needs to eventually hit 8K or 16K per eye at high framerates, which is not impossible for mobile chips if foveated rendering and other advancements are used. Pack it all in a monolithic 3D chip for a big boost in processing power.eva02langley - Tuesday, February 18, 2020 - link
Sorry to burst your bubble, but the maximum resolution of today consumer GPUs is 7680 x 4320 at 60Hz.Also, stacking dies for GPUs is not feasible unless you drop your power and frequency by at least half, which mean you would cut your performance at minimum by half of your silicon area. And if you stack 4+ dies together, you will drop the maximum performances per die even further due to heat which makes the whole concept totally non-viable financially. Basically, a double stack GPU would cost two time more and would barely provide 20-30% performance than a single die. Not to mention complexity in manufacturing such complex processors
edzieba - Tuesday, February 18, 2020 - link
"the maximum resolution of today consumer GPUs is 7680 x 4320 at 60Hz."For a single output link. For multiple composite links (e.g. DP MST) that can be easily exceeded, and for internal rendering (i.e. without the PHY limitation) it is VERY easily exceeded, limited only by raw ROP capacity.
nandnandnand - Tuesday, February 25, 2020 - link
Prepare to be shocked. Monolithic 3D chips are coming.yeeeeman - Wednesday, January 19, 2022 - link
it will take a looooong time to reach those resolutions. They will probably find other ways to make the image look sharper without increasing res.PeachNCream - Tuesday, February 18, 2020 - link
Based on the slow pace of product announcements and declining investment into R&D, it certainly looks like we are on the trailing end of our second VR fad. Maybe round three will finally bring us finely developed, useful products, but we still have not yet found broadly appealing use cases for teh technology yet so even if VR gets revived again in the future, it may still languish as a solution looking for a problem.edzieba - Tuesday, February 18, 2020 - link
HMD sales are continuing to ramp up rather than plateau.jordanclock - Tuesday, February 25, 2020 - link
Oh yeah, it definitely looks like VR is slowing.https://www.statista.com/statistics/671403/global-...
https://www.tomshardware.com/news/vr-ar-market-sal...
/s
PeachNCream - Tuesday, February 25, 2020 - link
Welcome to being seven days late to the party. In any event, you are presenting pull side data, not push side product launches and investment which is what I was talking about a week ago. If you look at historic data for new product launches, you will see that pull peaks always trail push peaks, but the curves are generally mirrors of one another.nandnandnand - Tuesday, February 25, 2020 - link
The article was nuked from AnandTech and completely inaccessible during those seven days.yeeeeman - Tuesday, February 18, 2020 - link
Hope this shit will have the same fate as 3D phones.nandnandnand - Tuesday, February 25, 2020 - link
Why? Would it make you even more antisocial?edzieba - Tuesday, February 18, 2020 - link
As far as I am aware, nobody has actually shipped a HMD based on the XR1 platform. Current non-devkit standalone platforms (e.g. Hololens, Oculus Quest and Go, HTC Vive Focus, Lenovo Mirage Solo) either use non-Qualcomm processors, or use use their own platform on a Snapdragon processor.valinor89 - Tuesday, February 18, 2020 - link
I have been searching and found the Vuzix M400, Glass Enterprise Edition 2 from google etc. They are not a consumer product, so I guess the XR1 platform was not exactly cheap to implement.PaulHoule - Tuesday, February 25, 2020 - link
Maybe it's just me, but "XR" has the stench of death about it."VR" hardware is "good enough" for many purposes but the content is not there; it will never be the time suck that browsing the web, watching TV, reading, or playing conventional video games because you can't stand long sessions. The hardware is expensive, but if there was enough content it might seem look a good investment.
"AR" is not good enough because of the problems that Karl Guttag writes about on his blog: the optics are a tough problem.
"XR" seems to have been invented to fit in somewhere between "The Race to 5G" and "The Race to AI". I imagine that executives in boardroom are asking "What is our XR strategy?" Just as the chemistry department birthed a materials science center, and then a nanotechnology center, a new name can convince (some) people that something new is going on.
nandnandnand - Tuesday, February 25, 2020 - link
If a company can make a wide field of view AR display that can be switched into a VR mode by putting a cover on it to block light, they should try it. There is already effort underway to flatten VR headsets and retinal projection could accomplish both AR and VR in a compact form factor.If it can be made indistinguishable from ordinary glasses or sunglasses (yes, including cameras), you can convince people to wear the AR device in public and switch it to VR mode when they are at home or somewhere they won't get mugged.
PaulHoule - Tuesday, February 25, 2020 - link
The "AR device that can be covered" will be a poorly performing VR device because the AR device has much more complex optics in the combiner.If you want to augment your world here and now, the birdbath architecture outperforms the diffraction waveguide at low cost. The only drawback is that it is ugly.
To get people excited about the platform you have to be able to demo it, show it off on a sales floor, put it on somebody's head at the art exhibit. The game developers might have three of them for a team of ten such as artist, coder, tester and want to be able to show the game off easily to visitors.
For that, you have to have a headset that works with or without glasses. That's the minimum viable product. Once people are sold on the platform, then the market opens up for small/light/personalized.
Alistair - Tuesday, February 25, 2020 - link
There's no point in calling the death of VR for a second time yet. We haven't even had major game companies go all in on it yet. We've yet to see PSVR2, or a Nintendo device, for example.PeachNCream - Tuesday, February 25, 2020 - link
At this point in the VR revival, that sort of thing really should have happened already in order to keep the industry healthy. It was in 2012 that Occulus first opened their Kickstarter which was, arguably, the beginning of enthusiam for a new generation of VR products. The lack of major effort by hardware manufacturers, console producers, and game developers across the board over the last half decad says quite a bit.PaulHoule - Tuesday, February 25, 2020 - link
I don't think VR is "dead" but I also don't think the problem with VR is with the displays. Tracking could be better and more universal. Thus no need for a PSVR2.Beat Saber is the Pong of VR. Somebody might put together a bestselling title if they can get the level of storytelling up to the level of Crazy Climber, never mind a Nep game. One of the most intriguing possibilities for VR is sharing a space with simulated characters, who can provoke strong feelings if they get too far from or too close to you.
If XR is the ecosystem of parts and systems software shared by AR and VR than it is a going concern. The silicon is mostly the same.
"XR" is the walking dead from a marketing perspective, however, because customers right now don't know what the value is in AR or VR and "XR" doesn't give any clarity.
nandnandnand - Tuesday, February 25, 2020 - link
So why did this article go dark for a week?The_Assimilator - Tuesday, February 25, 2020 - link
Someone unpublished it and forgot about it, because much like the VR market in general, it isn't important.