0
1

[–] TattedGuyser 0 points 1 points (+1|-0) ago 

I've always been fascinated with multiplayer games and the code that allows us to connect but I've never really had the time to delve too deep into it.

A topic that I often see discussed in around a few of my groups is the idea of server-side rendering for games, where only input is captured through the client and otherwise just acts as a receiver. For someone who has delved pretty deep into the topic (Your paper looks extensive, although I haven't had the chance to read it all yet), what are your thoughts on it?

0
1

[–] Alexaroth [S] 0 points 1 points (+1|-0) ago 

A big advantage I see with server/centralized rendering is that the server can also be incorporated in the rendering VM, which would make it the most authoritative server that can possibly be built (there should be no possible way of cheating against a setup like that). However, input delay now becomes the biggest issue, one that will not be able to be solved using conventional ways.

Unless you have VMs close to you, and you are paired with other players that are also close to those VMs, you will still see the effects of latency, more or less. So we're pretty much back to square 1.

I see some solutions being developed to solve that problem, but they are quite "brute force" in nature (rendering all possible next couple of frames achievable from a state and sending them to the player, so that no matter what the player does, he will be displayed the correct frame instantly, without delay). This is not a viable alternative, since it exponentially increases the workload of the rendering VM (even more when applying it to possible scenarios involving many other independent players connected to the same session, the more players, the more possible outcomes next rendered frames can have).

That being said, I personally believe that gameplay streaming will not help with masking latency in online games, at least not until we can figure out how to send information at faster-than-light speeds (which will solve the problem of latency in conventional client-server architectures anyway)

0
0

[–] TattedGuyser 0 points 0 points (+0|-0) ago 

Input delay has always been the biggest issue holding me back from accepting the idea of accepting server side rendering. Unless your a top tier company (Blizzard, EA, Ubisoft, etc), it would be extremely difficult/ near impossible to lay the infrastructure as close to everyone as possible to handle all the load. It would be interesting to see console manufacturers lay the foundation for what is essentially a 'user VM'. I suppose we still have a ways to go, great right up and hope to see more from you.

0
1

[–] illumnovic 0 points 1 points (+1|-0) ago 

I'm not OP, but I have been giving this a lot of thought lately. I think it's important to clarify first-off that server-side rendering usually doesn't mean that the dedicated server that is coordinating the clients does the rendering. Instead you have virtual instances from which you can stream games. I just wanted to preface that because gameservers generally lack decent GPUs to render games well.

That having been said, I think streaming games from dedicated virtual machines (VMs) is more an more plausible, especially in-home. Steam is doing a lot on this front and at this point allows you to easily play games remotely in your network. Problem is usually the connection speed, but especially over a wired connection this can work wenn enough. I've been thinking about doing this when I move in with my SO, because she's not a big games fan and it would put my louder equipement somwhere in the basement.

I think we're still a ways away from streaming from rented VMs though, because internet speed and reliability remain an issue. The other big issue is ownership. Already, when we buy games nowadays we usually only license the game, and therefore give up a lot of control over it, especially with DRM. This goes to the next level when you play on a VM owned and operated by the dev. It's hard to say how this will work out in the long run. I can see financial benefits for consumers who wouldn't need to build their own gaming PC, and it especially may be a model for consoles, but there are a lot of ifs and buts.