cancel
Showing results for 
Search instead for 
Did you mean: 

Is Rift destined to be a Windows only widget?

onefang
Level 3
From another thread, where it was off topic -

"haagch" wrote:
"onefang" wrote:
Please say it's impossible cybereality. I'm great at doing the impossible, once I know it's impossible. Oculus may think it's too hard, but I'm always up for a challenge. With Oculus lack of support for Linux, Mac, and laptops, I'll likely just have to write my own drivers. It will be completely open source, Oculus can switch to using it if they really care. 😜

You know, it likely is impossible. This part is the important one:
The current system has deep integration into the Nvidia/AMD drivers and Windows

Since oculus has chosen to "partner" with microsoft they have gone into a strong dependence into the proprietary ecosystem of windows and the amd and nvidia drivers.

Sure, you can always continue Doc Ok's work on reverse engineering and writing a new oculus rift driver from scratch http://doc-ok.org/?p=1095, but replicating and improving upon what oculus is releasing? Not without cooperation from microsoft, amd and nvidia. And then there's no guarantee that the CV1 can be reverse engineered the same way.


I'm beginning to think haagch is onto something here.

Normally I avoid Windows only widgets like the plague, since I'm not a Windows user. It's starting to feel like Oculus has not only suckered me into buying a Windows only widget, but the most expensive widget I've bought for years. I'm seriously rethinking this whole thing now.
81 REPLIES 81

martinP
Level 3
All this arguments that Linux is shell based OS unsuitable for VR is just ridiculous. If that was true I think Oculus would know that, at least John Carmack would for sure. And he would definitely share this information with Palmer so that he would not be commenting that "Linux support is on the roadmap post-launch"

haagch
Level 3
Let's look back at this:
"Tim74UK" wrote:
The exact same is true of any GUI interface that you may decide to use over unix / linux. The Gui interface is in effect a SHELL over the actual OS and addresses the hardware O/I via command line and back again...

not very efficient....

This inefficiency was the main reason Microsoft chose to make Window 95 directly access the hardware via drivers, they got the best of both world... Direct hardware access for efficiency, and an intuitive user friendly experience.

What does that mean, "addresses the hardware O/I via command line and back again"?

Have you heard of DRI/DRM?
https://en.wikipedia.org/wiki/Direct_Re ... astructure
provides access to DRM device
DRI3 revolves around using POSIX file descriptors for passing kernel objects between the display server and the application

How is that "via command line"?

And if you don't like that, have you heard of wayland?

It is exactly this reason that VR is not primarily suited to Unix / Linux... The whole reason this new VR revolution was able to exist was all to do with high throughput low latency hardware/software...


On the contrary.
The open and modular nature allows integrating VR rendering right into the desktop compositor instead of it having to sit "on top of it". Example: weston-rift: https://www.youtube.com/watch?v=3v7mOd0H06w
It also allows to extend the desktop compositor DIRECTLY for extended VR support. Example: The motorcar compositor: https://www.youtube.com/watch?v=Dgtba_GpG-U

the latency injected into the VR workflow via a gui shell running over Linux would be large and cause issues.

Please provide some evidence for that.

Could they get it to work... sure.... but why burn the man hours on something that is frankly used by such a minority of people??!??

You know, I have used the DK2 on Linux - not only that, I have used it with an intel/AMD radeon hybrid laptop with offloading to the radeon GPU. And once oculus released something that worked, the latency was okay. I'm not saying it was great, but it was okay. And I say that after having seen a couple of applications running on windows on high end hardware and as someone who is quite sensitive to stuttering.
Here, after 10 months (!!) oculus released a unity plugin that worked and I tried it and it was okay: https://www.youtube.com/watch?v=yIVlTGZv6WU. The Oculus World Demo latency tester showed 27 ms. Not great, but okay. The Readme said, the latency tester doesn't work correctly on mesa, but as I said, compared to what I have seen on high end pcs with windows, it was okay. And remember: This is on a platform that was severely neglected by oculus. I bet that it could have been much better if they had put a fraction of the work into it that they put into the windows version - where they still couldn't get laptop hybrid graphics to work properly.

Why burn the man hours? Indeed, why do it? Why did oculus advertise linux support on their website? Why did they not deliver in time even after getting 2billion dollar from facebook? Even if you think it's not worth it, they should have been able to use the facebook money to deliver on their promises! I.e. it shouldn't have taken 10 months to create a unity plugin, but 2-4 weeks at max.

The fact that you are willing to give up on the VR revolution because it isn't on your favorite Operating System of choice, is frankly, cutting your own nose off to spite your face!

Breaking Microsoft's quasi-monopoly and dependence of almost the entirety of several industries on a proprietary american (i.e. subject to the NSA NSA!) operating system is arguably more important than VR. Not everyone is willing to sacrifice that much freedom in regards to software for the convenience of VR. Go ahead with your "Social VR", spend time with other users. But be aware that you can only spend time in Oculus' version of VR with people who you can convince to replace their operating system with proprietary software from Microsoft and that the time you spend in this "social VR" you are actually cutting everyone out who is not willing to deal with microsoft.