Server based game augmentations. The transition to cloud. Really possible?

[Scott_Arm]

We dont talk much about Gaikai because we have seen it in action and there isnt much to talk about anymore until we see it in the PS4. Sony did not present it as something as big as MS's cloud based augmentations which is still "unknown" to us in practice. It is also implied that this feature will be part of the primary function of the console->game performance, gameplay opportunities and visual fidelity. It is supposed in theory to be used to cover the performance gap between the two consoles.
Gaikai is irrelevant to the normal gaming experience expected from the console: running a game locally like we always did for decades. MS wants to change that. Hence why we dont talk about Gaikai as much

Nevertheless, the infrastructure issues are exactly the same, but no one seems to be talking about it with respect to Gaikai.

I think a more interesting discussion is what could be done in the future with cloud computing assuming the infrastructure is in place. There is probably some value in putting bounds on the data that is being sent downstream to the client. The conversation is quite a bit different if you are sending 100k downstream vs 20M.

 

[Cjail]

@Nesh

What MS wants to do is a hybrid of cloud gaming, like Gaikai, and consoles so why you say Gaikai it's irrelevant?

 

[Boxercide]

Nevertheless, the infrastructure issues are exactly the same, but no one seems to be talking about it with respect to Gaikai.

I think a more interesting discussion is what could be done in the future with cloud computing assuming the infrastructure is in place. There is probably some value in putting bounds on the data that is being sent downstream to the client. The conversation is quite a bit different if you are sending 100k downstream vs 20M.

And yet we know Gakai and OnLive work. We've seen it. It's basically a streaming video service that accepts minute amounts of info (ie: controls).

 

[Scott_Arm]

For multiplayer games, that can be done on the server, so it's nothing new to cloud computing unless the algorithms are insanely complex needing super-computers.

I was listening to the podcast where they talked about how Halo 4 is using Microsoft Azure for the back-end, and one of the things they did was they wrote algorithms that detect griefers to automatically ban them. They can detect a griefer and ban them in 15 minutes. I didn't have a chance to listen to the whole thing, but there are a lot of sophisticated things going on behind the scenes to analyze data patterns to detect these kinds of things. I don't think a server provides you with the oversight needed to do that. You need all of the servers, or sessions (for peer to peer) dumping their data into a larger system with the smarts to analyze the data. It got me thinking about the idea of analyzing player metrics, and how that could be used to tailor gameplay.

For you idea of a game tuning gameplay, can't that also be done locally? Finish the game, and while you're reviewing gameplay stats or whatever, it cooks up some numbers that it feeds back into the game next time you play. Again, only impossible if the algorithms are insanely complex. Which makes sense . That why distributed computing exists, to deal with massively demanding jobs. A few stats analysis for single player hardly seems worthwhile. I suppose one advantage of the cloud is it can be working while the console is powered down.

Yes, it can be done locally, but like you said, you can handle large samples of the player population in the cloud, all of the time. If you want to tune things, or collect meaningful data that's probably a much better approach. It's far better than reading forum posts. I'm just throwing out ideas, but statistical analysis of player metrics to optimize gameplay is an interesting subject. I believe Bungie did this in a closed environment with some type of research center they have, run by a psychologist. They analyze and collect data from players in a closed environment. If you could get biometric feedback and player data from an online beta test, how much more useful would that be?

 

[Scott_Arm]

And yet we know Gakai and OnLive work. We've seen it. It's basically a streaming video service that accepts minute amounts of info (ie: controls).

Are you agreeing with me or disagreeing with me?

 

[sebbbi]

DF weights in.

Great article.

 

[Shifty Geezer]

I was listening to the podcast where they talked about how Halo 4 is using Microsoft Azure for the back-end, and one of the things they did was they wrote algorithms that detect griefers to automatically ban them. They can detect a griefer and ban them in 15 minutes. I didn't have a chance to listen to the whole thing, but there are a lot of sophisticated things going on behind the scenes to analyze data patterns to detect these kinds of things. I don't think a server provides you with the oversight needed to do that. You need all of the servers, or sessions (for peer to peer) dumping their data into a larger system with the smarts to analyze the data. It got me thinking about the idea of analyzing player metrics, and how that could be used to tailor gameplay.

AFAIK Uncharted has similar analysis. You only need game servers to report back to a development HQ. Cloud computing probably simplifies this a good deal and enables it for more developers, and I agree that's a good use, but that's not really the same as augmenting the game in terms of content, and isn't limited to the cloud. It's just one solution to the problem. I guess there's always more than one topology for any network based solution.

 

[Scott_Arm]

AFAIK Uncharted has similar analysis. You only need game servers to report back to a development HQ. Cloud computing probably simplifies this a good deal and enables it for more developers, and I agree that's a good use, but that's not really the same as augmenting the game in terms of content, and isn't limited to the cloud. It's just one solution to the problem. I guess there's always more than one topology for any network based solution.

What's the distinction between The Cloud and development HQ? Either way it is network based resources relative to the client.

The banning case is not an augmentation obviously, just an example of an algorithm that is running in the cloud that is analyzing player metrics for the entire population.

Why not use player metrics to tune gameplay? Send the player client new data for enemy stats in the background - data that has been tuned by analysis done in the cloud. Why is my level 40 mob frequently and easily killed by players 5 levels lower? Let the cloud tweak particular stats designated by the devs until the desired result is achieved.

Devs and publishers can't afford to build data centers. MS, Amazon can. They also provide the tools and environment to build these applications.

 

[Nesh]

@Nesh

What MS wants to do is a hybrid of cloud gaming, like Gaikai, and consoles so why you say Gaikai it's irrelevant?

Simply because one of the two is not a hybrid. If you can indeed call both "hybrid" they are in a different context. The DF article answers your question at the part where they described the difference between Onlive and what MS wants to do.

Gaikai was never brought up as a method to augment a game that you play locally. Similar to Onlive you just stream a video of game rendered somewhere else. Its a unique service on its own that can be optional and completely omitted without affecting how a game plays from disk or HDD. You put a game disk in? You will get a certain performance without relying on Gaikai

MS' though did present a hybrid solution where a certain game is supposed to be rendered using both your local hardware and the cloud to enhance your output and reduce the performance gap between the two consoles. And this involves actual game code. Not a live video stream of some game. It is supposed to be embodied with the primary use of the console. You put a game disk in the console? You expect cloud to assist how the game performs

We know what Gaikai does and its limitations and there isnt much more to talk about. You didnt see anyone talking about how Gaikai will make our games better after Sony's conference because it was never claimed to be what MS wants to do with the cloud

 

[Flux]

What are the tradeoffs with server/cloud based gaming vs console based gaming?

Are we just passing off the cost to the service provider to make the platform more accessible (cheaper for joe and jane smo)?

But seriously the latency is too high to be used for fluid gaming on a cloud.
The input and video lag is unplayable.(guess of 50ms-88ms best case average not counting random ping spikes is unacceptable)
Imagine trying to play a fighter or any other high paced game on a cloud.

Like I said. You cannot use a internet connection because of the inherent bandwidth and latency overhead.
I like the idea of cloud gaming because it will shutdown piracy and help independent developers sell their product in their own niche.

 

[astrograd]

Was this posted yet?

Xbox One cloud is 'one area console has advantage over PS4', Avalanche tech lead

Xbox One's cloud-side computing is one area that gives Microsoft's console an advantage over PlayStation 4, Avalanche Studios' chief technology officer Linus Blomberg has told VideoGamer.com, despite PS4 featuring more impressive 'raw power'.

Discussing the next-gen console alongside studio head Christofer Sundberg, Blomberg said that the console's ability to compute processes on the cloud is "definitely an area where the Xbox One has a step up over PS4, and it opens up for many interesting things.

"It's perfect for open-world games like ours as it enables techniques to make the game worlds more alive and social, such as persistence and asynchronous multiplayer features," he continued. "We are already using server-side computations in our PC hunting game theHunter, so that's nothing new for us as a concept. But Microsoft's solution may allow us to do this more efficiently and to a greater [extent]."

Xbox One's architecture is said to allow developers to off-load non-latency sensitive processes to the cloud, freeing up system resources on the console.

According to a report on Ars Technica, processes that could be off-loaded to the cloud include physics modelling, fluid dynamics and cloth modelling, each typically CPU-intensive.

A Microsoft representative (via stevivor.com) has also claimed that the cloud's processing capabilities will make the Xbox One "40 times" more powerful than the Xbox 360.

However, if the technology "proves successful for Microsoft", Blomberg expects "Sony to follow suit and add similar support" to PS4.

Whether Xbox One games perform will better than PS4 titles (or vice-versa), however, is "too early to say," Blomberg adds.

"I think [Xbox One and PS4] will be very similar in the end," Blomberg says. "Of course, if we find areas where either platform is particularly strong then we'll take advantage of that. But it's too early to say if our games will look better on any of the two platforms. All I can say is that there's potential for visually stunning games on both platforms."

Besides Xbox One's cloud-processing, Avalanche is also considering making of use of Kinect 2.0 and the console's TV functionality.

"If it makes the experience better, we’ll definitively support the Kinect and TV functionality," adds CCO Christofer Sundberg. "We have a few ideas at this point, but will only implement them if it adds to the core experience rather than take away from it."

Avalanche is currently gearing up for its "biggest" E3 since 2005, where it's expected to reveal a Mad Max title and a new game in the Just Cause series.

http://www.videogamer.com/xbox360/j...antage_over_ps4_says_avalanche_tech_lead.html

 

[astrograd]

Something I just thought of...couldn't MS gain a competitive advantage on their platform as a publisher by simply giving their first party titles more than 3 "units" of cloud computing resources? Just a random thought.

 

[Cyan]

This is the best article I've read on the cloud yet.

It compares the cloud to Xbox Live, Gaikai, Onlive, etc, and mentions exactly what differentiates them.

http://www.articlesnatch.com/Article/Video-Game-Evolution--Cloud-Gaming/4525124#.UaUkmjBBvDf

 

[XpiderMX]

Old?

http://www.eurogamer.net/articles/digitalfoundry-in-theory-can-xbox-one-cloud-transform-gaming

 

[patsu]

DF weights in.

Great article.

Yeah, the tech people here should be able to derive that level of info by themselves.

From what's been released/leaked so far, technically, it may not be very different from how existing MMOs split their work between the client and server.

What's more important is the business model. Microsoft will most likely provide a consistent and unified set of cloud APIs, and "free" servers for the games -- subjected to their terms. e.g., How much and who fund this ? It may be tied to the current debate about always-connected, in-game advertising, and game trading.

As for middleware providers, can they offer their physics++ engine as a common cloud service now ?

Next is what happens when the network failed ?

The low hanging fruits would be to allow Microsoft to serve subscription-based and free-to-play MMOs (and connected SPs) to consoles, phones and tablets. Games that are turn-base would be the easiest.

Without such "free" infrastructure, other companies will have to create their own environments in a vertical fashion.

This is what happened with Playstation Home. It kinda regressed into a vertical game platform with no/little relationship with other native games, The Home character management system can only be used by the PS Home titles. Initially, PS Home had some flavor of horizontal services for potentially all games.

In MS's case, the Avatars can probably be featured prominent in all cloud games.

 

[Boglin]

I'm all for Microsoft hosting the majority of mulitplayer games on their servers, as i'm sure it will provide for a much MUCH more stable connection for everyone playing. The only concern I have is for 1v1 games. Wouldn't a p2p connection be more ideal in a fighting game as far as latency goes?

 

[ERP]

I'm all for Microsoft hosting the majority of mulitplayer games on their servers, as i'm sure it will provide for a much MUCH more stable connection for everyone playing. The only concern I have is for 1v1 games. Wouldn't a p2p connection be more ideal in a fighting game as far as latency goes?

No because to send data from machine A to machine B and get a response you endup with 2x the number of machine to backbone hops and that is where almost all the latency is. Try doing a traceroute to google.com, almost all the latency is between your house and the ISP.
Connections on the backbone are by comparison very fast.

Also in a 1v1 game you'd want to provide a consistent experience for both players and if one machine is acting as a "server" that player would have less latency and most likely an advantage.

 

[astrograd]

The more I read of DF's article the more it seems misguided. They focus on bandwidth an awful lot early in the article as if MS is trying to replace internal buses with your internet connection. That's incredibly misleading, especially to make such suggestions at the forefront of an article like that.

Takes them forever to even get to stuff MS *actually* talked about. It's also odd they try to suggest X1's hardware wasn't designed for the cloud just because it "only" has LZ and jpeg compress/decompress in the DME's. Err...what more should they be expected to have in hardware for a software-based concept where the entire point is to send hardware-based computations outside the box? Hardware to help sending data out and in and...what else is needed?

Probably half of the article is written in a dismissive tone in the context of discussing the limitations of cloud streaming, which has little if anything to do with what MS is talking about. That's Sony/Gaikai's problem...not MS's. When they do talk about the claims MS actually made they frame them as if MS was "admitting" to something which is silly. When they talk about the specifics of latency insensitive cloud candidates they are also dismissive in tone and lump a bunch of pretty heavy computation tasks into vague categories and act like they are highly limited in nature without establishing why.

They also talk about the examples MS gave and then suggest MS is somehow light on specific ideas when those examples came during interviews with other publications. MS isn't going to give you exhaustive descriptions of every damn thing that is latency insensitive in a modern videogame. That's like accusing Bethesda of being 'light on ideas' simply because they don't talk up every single aspect of Skyrim in the very first interview on the game.

The most problematic part is that they seem to have missed the ENTIRE point, which is to use the cloud to free up local resources. DF doesn't just ignore this, they dismiss stuff like the lighting possibilities Matt Booty from MS mentioned simply because it could be done locally. Ummmm...that's the point! Moving stuff that is latency insensitive to the cloud to free up resources and get outta the way of the latency sensitive stuff. *sigh*

 

[Boxercide]

The more I read of DF's article the more it seems misguided. They focus on bandwidth an awful lot early in the article as if MS is trying to replace internal buses with your internet connection. That's incredibly misleading, especially to make such suggestions at the forefront of an article like that.

Takes them forever to even get to stuff MS *actually* talked about. It's also odd they try to suggest X1's hardware wasn't designed for the cloud just because it "only" has LZ and jpeg compress/decompress in the DME's. Err...what more should they be expected to have in hardware for a software-based concept where the entire point is to send hardware-based computations outside the box? Hardware to help sending data out and in and...what else is needed?

Probably half of the article is written in a dismissive tone in the context of discussing the limitations of cloud streaming, which has little if anything to do with what MS is talking about. That's Sony/Gaikai's problem...not MS's. When they do talk about the claims MS actually made they frame them as if MS was "admitting" to something which is silly. When they talk about the specifics of latency insensitive cloud candidates they are also dismissive in tone and lump a bunch of pretty heavy computation tasks into vague categories and act like they are highly limited in nature without establishing why.

They also talk about the examples MS gave and then suggest MS is somehow light on specific ideas when those examples came during interviews with other publications. MS isn't going to give you exhaustive descriptions of every damn thing that is latency insensitive in a modern videogame. That's like accusing Bethesda of being 'light on ideas' simply because they don't talk up every single aspect of Skyrim in the very first interview on the game.

The most problematic part is that they seem to have missed the ENTIRE point, which is to use the cloud to free up local resources. DF doesn't just ignore this, they dismiss stuff like the lighting possibilities Matt Booty from MS mentioned simply because it could be done locally. Ummmm...that's the point! Moving stuff that is latency insensitive to the cloud to free up resources and get outta the way of the latency sensitive stuff. *sigh*

You've actually missed the point and I'm quite surprised Shifty has been so patient with you.

 

[Boglin]

No because to send data from machine A to machine B and get a response you endup with 2x the number of machine to backbone hops and that is where almost all the latency is. Try doing a traceroute to google.com, almost all the latency is between your house and the ISP.
Connections on the backbone are by comparison very fast.

Also in a 1v1 game you'd want to provide a consistent experience for both players and if one machine is acting as a "server" that player would have less latency and most likely an advantage.

Okay. thanks for clearing it up for me Looking forward to some next gen Tekken haha