LinuxMCE Forums

General => Feature requests & roadmap => Topic started by: Veovis on August 30, 2009, 09:04:54 am

Title: Would it be possible to run a headless Windows gaming PC?
Post by: Veovis on August 30, 2009, 09:04:54 am
DISCLAIMER: I KNOW that this would be difficult.  I know that the devs here aren't being paid to do as I wish.  This is a hypothetical question, that, while I would be very happy to see come to fruition, I don't expect to.  I just thought this sounded cool and possible (not even remotely easy) at the same time, and wanted to share my idea.


I was looking at the OnLive (http://www.onlive.com/) service, and it gave me an idea.  What if in LMCE, one could hook up a high end PC to their Core, as well as maybe emulators running on either the core or the PC, and some real consoles, (Atari through 360) and play the games on any frontend they wanted?

Here's how I envision this working:

The video output of these consoles and PC all go in to the core.
All physical consoles are modded and hold all (legal) games in onboard storage, either first or third party.
The PC is running some simple software that lists all the games installed currently and launches them with any switches needed when talked to through LMCE's cross-program interface. (I forget the name.)
STEAM and other online distributors have a GUI on the frontends that lets people buy games from their couch.  Could be a source of revenue?
A games menu displays all games on all systems and lets the user choose any regardless of console.  Games on a console or PC already in use are greyed out.
The controller ports on the consoles lead into the core and handle input from the frontends.
The PC takes input through the core as well.
The PC either outputs a high resolution that the core downsizes for each TV, (better, because it allows "follow me" behavior) or the software interface to LMCE chooses resolution on game start based on the frontend's resolution.
Keyboard keys can be mapped to controller buttons and vice versa.

Please post any criticisms or additions to this theory here, or just talk about how cool it would be.
It is 1:00AM, so if anything is not clearly worded, please ask.
Title: Re: Would it be possible to run a headless Windows gaming PC?
Post by: tschak909 on August 30, 2009, 01:21:51 pm
In short, what you want is not possible, due to the fact that there is inherent latencies at virtually every stage of digitization and processing of the video signal.

The only way you can pull this off is literally with a multi-zone audio and video switching matrix.

-Thom
Title: Re: Would it be possible to run a headless Windows gaming PC?
Post by: Veovis on August 30, 2009, 05:52:59 pm
I've seen video of the OnLive service running over the internet, with minimal lag.  That's what made me think this was possible: if they can do it over the internet, I thought I could do it at home.  I'm signed up for the beta, as soon as they release it, I guess we'll see if theirs works as they've been showing.

Thank you for the response.

EDIT: I also forgot to mention, a few hours before I posted this idea, I was playing Fatal Labyrinth (a roguelike) on a genesis emulator on my computer, and to play it (since I can't find good 360 controller drivers that work with MacOS 10.6) I was using my iPod to VNC into my computer.  Moral of the story: turn-based games don't care about latency.
Title: Re: Would it be possible to run a headless Windows gaming PC?
Post by: colinjones on August 31, 2009, 01:17:43 am
For most games, however, latency is a key issue. Note that for the OnLive service to work correctly you need to be within 1000miles of one of their data centres, which is why they are building 5 data centres around the US. This is directly as a result of latency.

They are using a proprietary compression codec, which undoubtedly is intended to reduce the latency introduced by compressing the video, and they certainly would have very high-end dedicated compression front-end hardware in their data centres. Both these facts ensure that applying the same approach to your suggestion would be extremely difficult or impossible. Whilst the network latency on a LMCE network isn't a consideration, the latency introduced by compressing using an available codec, that is not optimised for low-latency on hardware that is not data-centre grade, would be a significant issue. Even only introducing a 1s delay between hitting a key and the reaction being displayed would make many games unusable.