At a first glance I do not see a reason why you would need extra computing power to add another headset. It is basically like a second screen which duplicates the image, so no extra power needed.
"Vega55" wrote: It is basically like a second screen which duplicates the image, so no extra power needed.
Duplicating the image to other person's HMD seems like a guaranteed VR sickness, eh?
Additional GPU is needed. Each HMD must have its own image rendered with its own position/orientation sampling provided for the game.
Answering OP: I see some factors which must be met in order for this to work: * Runtime supporting two HMDs at once on separate GPUs * Game supporting two HMDs and separately calculating all the data for them
These might be overcome as Linus did - by setting up a Linux distro with GPU passthrough, however this might introduce minor latency. Plus, the games would be completely separate from each other, meaning no local coop available.
1 - if it is simply duplicating then a single GPU (with 2 hdmi's) should be fine 2 - but if you want smth like in Linus show, where a PC was basically like a network workstation running multiple instances of a game then of course you will need to scale your PC power accordingly. But why?
"Vega55" wrote: Well there are two different approaches:
1 - if it is simply duplicating then a single GPU (with 2 hdmi's) should be fine
You wouldn't be simply duplicating as you are not going to be looking around identically to the other person and it has to render what is in your field of view so it is rendering 2 different scenes. (If you did just literally show what the other person is seeing then you would probably be sick in about 2 seconds.)
"Vega55" wrote: Well there are two different approaches:
1 - if it is simply duplicating then a single GPU (with 2 hdmi's) should be fine 2 - but if you want smth like in Linus show, where a PC was basically like a network workstation running multiple instances of a game then of course you will need to scale your PC power accordingly. But why?
Because 1) would be painful for the person who receives the duplicated image. You basically would see what someone else sees, however including their head movements and rotation, which is different from your position and orientation. A definitive motion sickness, if you ask me.
In order to share the game between two people, you must redraw the picture twice - once for player, once for spectator. This is only because you need to separate spectator's head movements from player's head movement. And for that, you need another GPU (plus Runtime/SDK and Game support for second HMD, but that's another story).
I don't see the need for this other than to just do it cause it's fun and cause you can say you did it.
I love Linus Tech don't get me wrong. But really I can't see any advantage to doing the above over buying seven gaming pc's, other than perhaps power consumption and physical space requirements.
In fact I can see a whole host of reasons (no pun intended) to not do the above, such as if there is a hardware issue on one component it effects all etc...
I was thinking that quite a few will be building monster builds anyway with 2 GPUs for SLI or whatever.
If you could do some shared experiences on the one machine like the touch demo type think or 3d painting or just 2 players playing a racing game or something.
Linus essentially runs 8 VMs on 1 rig, therefore nothing stops you from running 2 VMs on a rig that can pull it off with VR. Basically you need a mobo with 6 usb3.0 ports + 2 usb2.0 ports exclusively for Rifts, plus as many USB's as you need for other peripherals like mice and keyboards and USB sound cards - everyone wants to have sound in their game, right?
In E:D it would be a fantastic feature to fly as a team through space while sitting next to each other irl, using a single PC. Sure, it sounds like a great idea, but hardware isn't quite there yet. Give it another generation or two.
Not a Rift fanboi. Not a Vive fanboi. I'm a VR fanboi. Get it straight.
