Fool me, this is real good shit ! [the VR thread]

[Dvorak]

Ordered.
Let's see.

[Jamesac]

The new version has massive improvements to motion blur and adds head tracking. Can it be the future yet?



James

[Deckard]

Of all the tech in the last few years this is the one I feel most impatient for - Any ideas when we might see a release ?

[Dio]

Guru words on VR - Mike Abrash weighs in.

I want this to be huge for at least two reasons (it sounds terribly exciting, and the graphics horsepower requirements should guarantee me meaningful employment until retirement age ).

[Mqark]

I was a little gutted to hear his expectation that it'll be 2015 or a little after, however to get the best possible it may be worth the wait rather than jump on-board only to have to replace a significant investment the next year.

From what I've seen so far (the HD Oculus (not Crystal Cove)) it's pretty darn good already, but if the "Presence" that he's talking about is attained, then this is going to be stonking. I'd love to give the Valve headset a go, but I suspect only serious developers will ever get their hands on that.

My only requirement so far is greater resolution to eradicate the still noticeable (and a little off-putting) screen-door effect, the 3D is already the most believable I've ever experienced. I had no problem with the persistence of the pixels causing issues.

People shouldn't forget just how significant a leap that even the worst 3D that we can muster nowadays is, It's significantly better, almost immeasurably better, than the 1990's 'Money for Nothing' slow,flat shaded, low polygon count, filled poly objects that required quite a suspension of disbelief to accept that you may be in a 'virtual world'.

[gravy]

Cheers for that Dio, excellent read

[Dio]

The one thing that really surprised me was the low persistence requirements and then the corollary of that being the very high framerate requirements.

There's going to be a lot of data stuffed down that cable .

[zip accident]

A couple of thoughts on what is a great read.

One: two years? he's talking about an equivalent of the two years from the first Whittle jet to the first flight of the Gloster Meteor, which was only capable of 400 mph, later versions of the Meteor got up to 600 mph, but the first was not that great, though it was a sign of things to come (the Germans started in about 1938, and had the better but not perfect Me 262 flying by 1944).

Second: PCs are where the FLOPs are? that's a huge maybe unintentional swipe at AMD's Bulldozer based systems with Floating point units shared between two integer cores. If it's Flops that are needed then maybe using a second GPU for compute would be in order?

Third thought, some of the aberations in lenses he talks about are spherical aberations, which can theoretically be avoided by using parabolic lenses, though parabolic lenses are much harder to creat than spherical ones.

[Dio]

Second: PCs are where the FLOPs are? that's a huge maybe unintentional swipe at AMD's Bulldozer based systems with Floating point units shared between two integer cores. If it's Flops that are needed then maybe using a second GPU for compute would be in order?

He means because the GPUs in the consoles are equivalent to lower-end-of-mid range PC parts. You get an order of magnitude more graphics horsepower if you're prepared to pay about the same for a video card as you do for a whole console, and feed it the watts that you need to render quality graphics 100 times per second at 4K resolutions.

[gravy]

The one thing that really surprised me was the low persistence requirements and then the corollary of that being the very high framerate requirements.

There's going to be a lot of data stuffed down that cable .

Yeah You're right, this takes off and the GPU arms race where the end user can really appreciate the end results is back on.

[Dio]

It is going to be an interesting challenge to sell the 4K plus ones. Could be talking a $500 peripheral that needs a $500 video card .

[Misfit]

It is going to be an interesting challenge to sell the 4K plus ones. Could be talking a $500 peripheral that needs a $500 video card .

Not that hard...

So, as someone thinking about building a gaming rig at the moment. Is a GTX 780 going to be enough to keeping me going in this brave new world?

[zip accident]

Is a GTX 780 going to be enough to keeping me going in this brave new world?

Given that will be a two year old card by the time this arrives, possibly not. If you want to build a PC for this upcoming VR situation, it's probably best to wait until we know what the specs are. On the other hand, it's a pretty strong card now, and it might be barely enough then.

OTOH if you're building a PC now for now, it's a strong card, and there're probably not many games to stretch it at this point, or its AMD equivalent. I think if I was trying to build a humdinging machine, I'd probably be looking at the twelve core/24 thread Xeons, but unless I was trying to drive a 2560x1600 display (or a multi-monitor set-up) I wouldn't be worried about the capabilities of a GTX 780 being too slight.

[Dio]

There's no need to go too silly on the CPU side. CPU requirements are going to be set by the consoles and their relatively modest clock rates mean a PC chip doesn't have to be too crazy. And with Mantle in the works, the need for a big excess of CPU just to keep up with a console is going to recede.

I make my living room machines do me for about five years (with a video card bump - easy for me! - and maybe some more memory half way down the line) and I just finished building one that I expect to do me very nicely for the first round of VR, using the top-spec 4.1GHz A10 with a 6850, 8GB RAM and a 256KB SSD. The 6850 is from my spares pile, admittedly, and I reused the old case (new PSU though), but the rest of the machine was £400 and not far off a third of that was the SSD. It's a phenomenal piece of kit, nearly silent, even with intensive 3D games running. A big improvement.

VR is going to be all about graphics muscle. Even just 1080p at 100fps with all the visual trimmings is going to call out for every transistor on the top R9s, because you want that 100fps guaranteed in the worst case scenes - you don't want to be kicked out of immersion just because it's getting busy.

[Whiteball]

I have no idea about these types of things but, what about a variable sync rate using a G-Sync or 'free-sync' type of thing?

Would that make any difference to the workload or any latency issues etc?