Looking Glass needs your help!

This is a call out for help from the community.

I have started to update the NvFBC capture code in Looking Glass (http://looking-glass.hostfission.com) and in theory it should be working provided you compile the host application with the NvFBC API SDK. However as it has been pointed out several times in the past, NVidia do not allow consumer cards to use this API (unless you’re Valve or ShadowPlay).

It is known that there is a way to make it work on consumer cards, however as I am bound by the API’s License agreement use of these methods would land me in legal jeopardy and possibly also this project.

I am looking for anyone that would be willing to either donate the card itself or the funds to purchase a NVIDIA Quadro P5000 or better. This is not a cheap card as it’s a “Professional” device coming in at $2900 AUD.

Having this device would allow continued development on the NvFBC capture interface for those that have professional cards, or those using special methods to enable the API on consumer cards (which I can not officially condone).

I have setup a GoFundMe that will cover the cost of this purchase for those that which to donate financially. Any excess that is not used will be held for future purchases for the project should the need arise.

Thank you all for your support!

8 Likes

I may be able to get you a gently used one for around $1k us or so, will see what I can do

7 Likes

I didn’t know that you can enable NvFBC on consumer cards, if it is not legal, it means that I can void warranty by doing it?

1 Like

No, it will not void your warranty, but I can not release a version of LG that has the magic in it to make it work as I would be in direct violation of the SDK License Agreement, nor can I use it myself as it would taint LG as a project.

3 Likes

So if I enable it on my GPU (GTX 1060 - if is it possible), then I can use NvFBC with Looking Glass? :slight_smile:

I can’t advise on this, but I can say it’s not something you enable, it’s the software you run that has to “enable” it for it’s exclusive use.

1 Like

So and what is real advantage of it? I know that Looking Glass uses Windows Capture API, so it means that by using NvFBC we will get lower latency?

Are there any measures, what we can expect? :slight_smile:

From what testing I have been able to perform I can confirm that it breaks the 1200p resolution limitation. The NvFBC capture API is far superior to DXGI.

2 Likes

Remember, he needs to keep out of trouble. Nvidia’s agreement; he has no choice but to abide with it. No different to other IT-related projects of a bigger scale than this.

But it will be an interesting mission to find a Quadro or better. Makes you wonder what other stuff @gnif can use the card for once he gets the project sorted out.

3 Likes

Honestly a GRID would be more interesting, but I feel thats just asking the community for too much.

I’ve recently been realizing, as I continue down this Linux path, just how much I rely on other people solving really niche problems for me. I’ll gladly give what I can and thanks for all the effort you put into this!

2 Likes

@gnif there are plenty of P5000’s available through Paperspace for $0.75c an hour, if you only care about developing the capture method that could be a good place to start, but obviously aren’t going to offer an end to end solution. In addition, if you just care about testing… you probably can start with a P2000, or M2000. They’re the lowest end VFIO cards with driver support for PCI Passthrough (I have an M2000 in my test rig) - and a LOT cheaper…

In addition, the GRID/Tesla stuff is extra painful because of the licensing, with the enterprise license with grid…you basically are just paying to unlock the Quadro features…so just starting with a Quadro is the right method and will not have any licensing difficulties.

Thanks @jamesstringerparsec for the information, unfortunately it’s not simply enough to rent access to one, I need one physically on site as there is no way I can profile or evaluate it’s performance for low latency usage when the system is in a remote location.

As for the lower end ones, yes they would be viable however I need to have one capable of running some of the more recent AAA titles at 4K resolutions, or high refresh rates for 140Hz and 200Hz monitors, as this is where LG is suffering the most at current.

You’re likely right re GRID licensing, I have no official experience with GRID at this point other then hacking GTX960’s into them several years back.

Want to borrow a Tesla v100 32gb if it’s useful?

2 Likes

I’d love to play with one but I don’t think it’s going to be useful when in comes to Looking Glass. You would need some way to convince it to act like a normal card with a display attached.

Are you planning something similar with Radeon cards ? I’m planning to stop buying nVidia after the error 43 crap. Yes, I can work around it, but I’m not cool with that.

AMD cards already work using DXGI Desktop Duplication, and as far as I am aware AMD deprecated their propriatary capture API in favour of a good DXGI Desktop Duplication implementation.

A post was split to a new topic: Is AMD DXGI Superior to NvFBC

Sorry for resurrecting an old thread, but would this work with a Tesla A100 with the “MIG” technology?