Why support VR on phones/GearVR

[Warren989]

When it will most definitely fail as a viable platform.

I mean honestly, all this talk of cells replacing computers is ridiculous.

Cell phones will never catch up to the power of a computer, ever (maybe graphene/qubit CPUS). There's nothing wrong with that, I mean, they were never suppose to.

Lets remember that ARM stands for "Advanced RISC Machine" and "RISC" stands for "Reduced Instruction Set Computing". RISC works well when you have a very specific purpose in mind. That's why they are used on stuff like cells, embedded devices, and bit coin mining. RISC chips are set to run a specific set of instructions over and over again, and so clock for clock can destroy even a big boy like the 5960X if running very basic code ad infinitum.

Most code only requires a few instructions to run — read/write memory, do arithmetic, jump, boolean logic, not much more. ARM processors only offer these basic instructions at speeds that are acceptable in small scope as they are integer based. A significant amount of code running on a PC is floating point, not integer.

But that's where things get messy.

X86/X64 is actually faster AND more power efficient when dealing with more complex code as they are "CISC" or "Complex Instruction Set Computing". Although almost all code can be represented by the basic instructions in RISC, certain patterns of instructions are common — for instance, “write this byte to memory then look at the immediately following byte”. CISC processors offer combo-instructions (example being STOSB) that handle these common instruction patterns super-efficiently. However, support for these combo-instructions requires extra hardware — and that hardware costs electricity.

That difference in hardware is why ARM processors use less power than x86/x64 processors at the same clock speed. Mind you, it also means that some programs run faster on x86/x64 processors than they do on ARM processors with the same specs — an algorithm that takes 3 cycles on an ARM processor can take 1 cycle on an x86/x64 processor if it’s been wrapped into a CISC combo-instruction. It’s also why programs built for x86/x64 can’t run in ARM — once you compile a program for x86/x64, it’s hardcoded to use these combo-instructions, and there’s no translation to ARM from there.

See where I'm going here?

There's a reason almost all cell phone programs are basic, very basic compared to their PC counterpart. The more complex the the software is, the slower it runs on an ARM. I mean, that's true for any CPU, but even more so for an ARM since it does not do CISC at all, and only offers limited FP (Floating Points). Even the newest ARM CPUs trail that of an Intel Atom in Floating points which is not saying much as an Atom is beyond slow. The PC world is moving to a more unified GPGPU environment which is all FP, not Integer calculation which ARMS do best.

Now one might say, why not just use Int? (Integer). The problem is Int is fast for a reason, there's less calculations as the value is always true or false and it's in binary. Example Int is:

game_started = 1
if game_started == 1 then.....

Versus Floating point:
game_started = true
if game_started == true then...

See how running Int code could be a problem making more complex software/games for a mobile VR?

It could be done, but it would remove the point of ARM/Int which is short simple code that does not require dynamism.
There's a reason we all don't just program in binary even though is much faster than FP, and that's due to complexity and dynamism.

This performance is NOT acceptable for VR:
http://www.slideshare.net/napoleaninlon ... omparisons


Its a long paper but if you scroll to the end you'll find a table of benchmark results and some analysis. The raw figures is below:

Dhrystone(DMIPS) Whetstone(MIPS) linpack(KFlops)
ARM A8 Cortex 883 100 23376
ATOM N330 1822 1667 933638

What can a full on x86 core do, ie a 2600k? Well http://www.guru3d.com/articles_pages/co ... ew,13.html breaks out the dhrystone and whetstone results. The results on guru3d are in Giga ips so they are a 3 orders of magnitude off, so I've converted them to MIPS below:

2600k 118000 83000 ~40000000

A 2600K which is already a few years old is basically 1000's of times faster in synthetic benchmarks. These results aren't directly comparable because Sandra Dhrystone and Whetstone are different beats to the benchmarks linked but if you run linux you can quite happily download and run them yourself and see how quick your system is in comparison. The figures are orders of magnitude however so no amount of messing with the benchmark is going to stop them being many many times faster.

In short there is a massive difference. And trying to get something as complex as VR running on a cell phone is not only silly but risks ruining support for the masses once this endeavor inevitably fails miserably. People will associate VR as a fad once it fails on phones/consoles and perception/confidence will falter affecting support on any PC VR solution.

The one last thing I'll say is this:
Does the GearVR demographic even care? Even my GF and all her friends only find the DK2 amusing at best, just another toy at worst. PC fans are the ones which would likely support/buy a VR solution in the first place as we are the ones with the Rigs, forethought, and bucks to support it in it's infancy.

So who the hell is this even marketed towards?

Casual people who imgur and candy crush will not be buying into VR. How can FB not know this?
I know games/tech is not their market so I can forgive stupidity on that front, but still, this should be blindingly obvious to people in whatever marketing department they have.

I really hope Oculus gets back to what made it, PC fans who want a true VR experience and not a 3D cinema or some gimmicky virtual chat room which is about the best a phone could handle.

[cybereality]

Sorry you feel that way.

Have you tried the Gear VR yourself? Cause it's probably better than you think.

[EarlGrey]

If you can get the experiences you want on mobile then it doesn't matter which one is faster.

You can run 3D games on mobile which required high-end PC rigs just 6 years ago. In another 3-5 years will you be able to run on mobile today's AAA games? Most likely!

If you want bleeding edge graphics, yes, PC is for you, but for most mobile will suffice in providing them the experiences they want with the convenience of mobile.

Also there's a big waste in PC's, they're not as optimized nor is the software, as mobile hardware and software is. There will be demand for smooth VR on mobile and rich experience, and that's going to be possible (and already is on GearVR)!

So the debate isn't about whether PC is faster than mobile, of course it is, the debate is... can you get good enough experience on mobile? The answer is yes.

[Warren989]

"EarlGrey" wrote:

You can run 3D games on mobile which required high-end PC rigs just 6 years ago. In another 3-5 years will you be able to run on mobile today's AAA games? Most likely!

Can you name a game from 6 years ago that was demanding that runs on a cell?
You do realize that to run a game like BF4 the ARM CPU's would literally need to pack 2-3 billion transistors, right?
Not gonna happen.

"EarlGrey" wrote:

If you want bleeding edge graphics, yes, PC is for you, but for most mobile will suffice in providing them the experiences they want with the convenience of mobile.

That's my point. No one will have this strapped to their head on the bus, or walk around the mall with this. Wrong target audience.

"EarlGrey" wrote:

Also there's a big waste in PC's, they're not as optimized nor is the software, as mobile hardware and software is. There will be demand for smooth VR on mobile and rich experience, and that's going to be possible (and already is on GearVR)!

That's a myth. It's like when people say consoles are more "optimized" code for we cut corners. Cell software is Integer, because the ARM chips are Integer as they are RISC which limits their complexity. Did you even read my post?

"EarlGrey" wrote:

So the debate isn't about whether PC is faster than mobile, of course it is, the debate is... can you get good enough experience on mobile? The answer is yes.

I doubt that. You just won't ever be able to play more than a very simple demo at most as the hardware AND software just isn't there, limited by it's nature of being a low power device.

[geekmaster]

Some people will never be happy, even when they have direct neural interfaces. Such is their nature, and it permeates their outlook on life and their basic perception. Personally, I like to look on the bright side of life, to the best possible futures, and to help them come into fruition. Pissing on somebody else's meal is easier than cooking your own, for such folks. I am glad to be who I am, and to create the futures that I dream. May we all have such joy in life.

Personally, I enjoy my GearVR, and much of the complaints I hear about (and even what John Carmack said in his presentation that we "must learn to live with") have obvious (to me) solutions on existing hardware. But I will need to provide working demos to convince most folks, and I no longer feel welcome sharing the fruits of my research in this forum. I will need to find a new place to post where I can delete posts and ban folks I find toxic to my own peace of mind.

[Warren989]

"cybereality" wrote:
Sorry you feel that way.

Have you tried the Gear VR yourself? Cause it's probably better than you think.

I have not tried it. However this is more about the technical limitations as to why this is a bad idea. And those hold true regardless of personal opinion. That's why I listed technical limitations vs just saying it sucks because PC master race.

I'm sure quite a few people will find cool. But is that audience big enough? A product like this really only appeals to casual users, due to the low graphical fidelity and that the content will be inherently limited by any SOC (embedded) device. Casual users are fickle, they tend to find things like this more of a novelty than an experience.

I'm not saying this is a bad product, just a bad execution.The target audience is backwards. You build from your most fervent fan base then move outward, not the other way around.

[EarlGrey]

I once did this algorithm, it took 100 ms. to run, totally unsatisfactory.
I refactored it, and managed to get it to 20 ms. Still slow.
Then I used better data structures, got it down to 5 ms. Pretty good.
Then I decided to cheat a little, try to put some more knowledge into the algorithm, got it down to 1-2 ms.


Software is what runs on hardware. If the software not optimized well enough it will run slowly no matter how powerful your hardware is.

I've seen video comparisons of games running on a high-end PC vs. optimized for a PS3 or Xbox. And honestly it was almost impossible to perceive any difference in quality, despite the consoles having lower resolution and less detail. You needed to have the video playing side-by-side comparison at slow speeds to be able to see a tiny negligible difference.


It's the tethering that kills PC's. If you could stream the video to the headset with no latency then the PC would be a clear winner. But mobile seems to be an obvious consumer choice.

But you need to stop looking at this from a technical perspective and look at it from a human perspective. Technology should fulfill human requirements, not the other way around, and human requirements are that VR is tetherless.

[Warren989]

"geekmaster" wrote:
Some people will never be happy, even when they have direct neural interfaces. Such is their nature, and it permeates their outlook on life and their basic perception. Personally, I like to look on the bright side of life, to the best possible futures, and to help them come into fruition. Pissing on somebody else's meal is easier than cooking your own, for such folks. I am glad to be who I am, and to create the futures that I dream. May we all have such joy in life.

Personally, I enjoy my GearVR, and much of the complaints I hear about (and even what John Carmack said in his presentation that we "must learn to live with") have obvious (to me) solutions on existing hardware. But I will need to provide working demos to convince most folks, and I no longer feel welcome sharing the fruits of my research in this forum. I will need to find a new place to post where I can delete posts and ban folks I find toxic to my own peace of mind.

I have no idea if this was directed at me or not, little generalistic. I'm not pessimistic. I listed technical limitations as to why this is a bad idea. I've read your posts, and I would have thought that someone as technically minded as you would acknowledge the limitations/user base and understand why this is not the best approach.

I love the people behind this company even more than valve. They are more of a faceless company then Oculus was. But lets remember, Oculus is owned by FB. So calls are made from them, not Oculus, not Palmer, not Carmack.

And even if they were, a bad business choice is still a bad business choice, regardless of who makes it.

I'm not blinded by emotion. I separate the decisions being made from the people working there.
And to focus on mobile is a bad decision, for now.

[EarlGrey]

"Warren989" wrote:
And to focus on mobile is a bad decision, for now.

To not focus on mobile would be a horribly bad business decision. You'll have 50 million mobile VR users long before you have 10 million PC VR users.

But that doesn't mean it's an exclusive focus. You know companies like Oculus have more than one people working there, many groups, departments. Not all have to be focused on the same thing. And no, Oculus focusing on mobile VR doesn't mean it's not focusing on PC VR as well.

This has nothing to do with VR mobile being practical, it's about your perception that it has an effect on PC VR. You realize that advances in mobile have made VR a reality for PC's? Tech advances in mobile carry over to PC. To be against mobile VR might end up hurting PC VR.

If mobile VR becomes a huge hit, that means more R&D towards better panels for VR, which means PC VR benefits.

[PeerReview]

What kind of media will be available on mobile? As soon as the platform is there does that mean developers, in particular AAA developers will start making media to be consumed on it? I am all for indie content don't get me wrong but for this to be viable, and capable of actually making serious revenue there will need to be some AAA quality games and extremely good support for Hollywood films. Although what I personally would be most excited for would be software that is different than anything available. Like getting ready to take a job in another country and being able to walk around the neighborhood where I'm moving, or taking a tour of the new facility and my lab. Perhaps letting my 75 year old dad walk around Newport Beach again like when he lived there in the 70's. Or countless other things all the creative people will come up with.

I guess what I'm getting at is, what are they looking to do with it? What are the use cases? Also what other solutions are there for the stereoscopic 3D on mobile platforms besides some kind of OpenGL wrapper? I think this could potentially be great and am excited about it, but I have so many question, and even more concerns. If someone could shed some light on what exactly is to be accomplished with this I would be extremely grateful. Thanks in advance for any replies.