OnLive: Cloud-Based Gaming of the Future

[Andy Chalk]

OnLive: Cloud-Based Gaming of the Future

A new system called cloud computing [http://www.onlive.com/]" to deliver on-demand gaming without the need for expensive console and PC hardware and upgrades.

Developed over a seven year process by Steve Perlman and Mike McGarvey, the Onlive service works by handing off the work of running the games to high-end servers, which handle rendering, AI and other gameplay issues separately from in-home hardware. User input is transmitted to the servers while video is streamed back over broadband networks to customers who can access the service through conventional PCs including laptops and netbooks, or with a "MicroConsole" provided by OnLive. As a result, the need for high-end gaming hardware, even for resource hogs like Crysis [http://games.ea.com/crysis/], is eliminated.

"This is the last major console cycle," Perlman said. "If not this one, then definitely the next one."

Keeping things simple is the client software, which offers the same interface across all platforms. The goal is to have games on the service load "nearly instantaneously," and Perlman claims the latency is at least as good as, and usually better than, playing on a LAN. OnLive works thanks to a proprietary system of on-the-fly video compression which will require a minimum 1.5 megabit connection for standard definition (480p) and a five meg connection for hi-def (720p or 1080i) resolutions. Support for 1080p and higher resolutions is expected in the future.

OnLive has already attracted several mainstream publishers to its platform, including Eidos [http://www.ea.com] and others, offering benefits including a simplified development process, reduced production costs and of course the virtual elimination of piracy. For their part, users get simple, on-demand gaming they can access without the need for costly hardware, although hardcore gamers may be put off by the loss of things like mods and performance tweaks.

Despite the optimism, there are downsides, the most obvious being the need for a reliable fat pipe in order for the thing to work. Questions about privacy are unavoidable, as are concerns about the loss of games and data should the service fail. There's also the healthy skepticism that follows (or at least should follow) every "too good to be true" story, which this certainly qualifies for, particularly in the minds of those who remember PlayStation Network [http://en.wikipedia.org/wiki/The_Phantom_(game_system)], which themselves represent a half-step toward an OnLive-style environment, it's hard to argue that this service or something very much like it won't loom large in the future of gaming.

Source: ExtremeTech [http://www.extremetech.com/article2/0%2C2845%2C2343703%2C00.asp]


Permalink

 

[KDR_11k]

AHAHAHAHAHAHAHA

Thin clients suck even when you just run a standard productivity app on a local server, a videogame over the internet is going to suck so hard you can use it to clean your room.

Thin clients ALWAYS come with promises like "you'll never have to upgrade your hardware again!" and then users complain about the horrible performance compared to a conventional thick client and in the end the thin client usually gets abolished. I have no idea how often that scenario has been dealt with in the BOfH stories but I've seen it happen in real life too.

Hell, how are they going to fix the lag issues? Rewrite the laws of physics? Sorry but signals only propagate so fast. Add the overhead from all the hardware in between and the theoretical minimum latency isn't going to be pretty, never mind the real life situations.

Oh and don't forget about ISPs deciding they don't like how much bandwidth you're wasting...

 

[Lord_Ascendant]

So its Steam meets Netflix?

 

[Ranooth]

Said it before but this be more offical. If they can get a good amount of support behind it then it will change a lot of things.

Interesting to see where we are in a years time with this.

 

[fix-the-spade]

Sounds great in theory, but I can already hear the phrase subscription based service being bandied around in the future.

No doubt when this system comes into place it's going to be heinously expensive and work nothing like the Steam Cloud.

 

[plastic_window]

I really like the sound of this - if they get it to properly work and if they can sell it without making it too expensive, then it will have a massive impact on gaming. Fanboys will be in short supply, the graphical quality of games can flourish if everyone can play Crysis on their PC - meaning the graphics whores can't pretend to be superior for having a bigger rig than me.

It if fails, it'll just fall into obscurity and in 3 weeks no-one will care. I have a very good internet connection, so I'm not worried about the lag, but it would suck for people who can't enjoy HD gaming because where they live doesn't support broadband.

The way I see it, either outcome is fine by me - though obviously, I'd much prefer the one where it works. It sounds almost too good to be true.

 

[TheBluesader]

I really like the sound of this - if they get it to properly work and if they can sell it without making it too expensive, then it will have a massive impact on gaming. Fanboys will be in short supply, the graphical quality of games can flourish if everyone can play Crysis on their PC - meaning the graphics whores can't pretend to be superior for having a bigger rig than me.

It if fails, it'll just fall into obscurity and in 3 weeks no-one will care. I have a very good internet connection, so I'm not worried about the lag, but it would suck for people who can't enjoy HD gaming because where they live doesn't support broadband.

The way I see it, either outcome is fine by me - though obviously, I'd much prefer the one where it works. It sounds almost too good to be true.

True, true, true. Some of us have crappy old PCs but really cutting-edge broadband, and it's cool how even crappy old PCs can handle a fast connection. If I could use that connection to bypass the need for, say, graphics processing, I could see great images in my browser that I could never "create" with my own computer. And as long as broadband speed keeps its toe in old computers, like it is now, this will continue to work.

Let's see where this goes, of course. But if Steam functions properly in reality, anything is possible.

 

[FistsOfTinsel]

The only people who would think this would be superior, lag-wise, to current online gaming are those who have never played games online. Sure, your shots won't lag behind your target; instead the whole world will lag behind your controller. At least you won't have to worry about unfairness in terms of who got shot - everyone's experience will be as shitty as your own. Maybe ten or twenty years down the road the net infrastructure might be such that you will be able to play with only a few milliseconds of screen lag, but not today. I'd say it's at least 3 console generations away.

It's obvious this has absolutely nothing to do with saving users (or developers) the "hassle and expense" of upgrading hardware (the cost of your console is small compared to what you spend on games, DLC and Live memberships), and everything to do with guaranteeing that every soul playing your game is a playing customer. Not only do you have no more pirates, you have no more "leeches" denying you cash by purchasing the dreaded used games. Never mind the fact that used-game gamers (like myself) are essentially subsidizing those who choose to run out and buy the games on release day.

This is the holy grail for publishers - there is absolutely, positively no way for someone to play a game published this way without paying the publisher. Even getting the source code and hosting their own server would require effort way beyond what your typical downloader, pirate or used gamer puts into getting their goods.

 

[phirewind]

one word: idiots

"this is the last major console cycle"... yeah, and nobody will ever need more than 640kb of RAM. Think about areas where the provider has a 50GB monthly cap on internet customers. That's 22 hours and 45 minutes of 720p gameplay per month. How about this, I run it on my own computer and play at 1920x1200 as much as I want for free?

Secondly, how can this possibly be feasible with new game releases? Unless the games are set to run at "absolute crap" visual quality, it still takes a lot of power to render at 30fps at 1280x720 resolution, not to mention compress the live video stream. What business model could possibly make money where each end user requires 5mbps downstream bandwidth + several hundred dollars of dedicated hardware unless it's a long-term monthly subscription + cost of the games? Then you just pray that no more than X percentage of clients ever decide to play at the same time? Oh no, Crysis 2 just came out, every client is trying to play it, servers are hosed! Now you bought the game and won't be able to play when you want until the service provider upgrades, or other people get tired of playing. I can see where this might have a real place in the market in 4 or 5 years, but right now the USA internet backbones just won't support the customers for it at any decent scale.

Xbox Live and PSN are nothing like OnLive, not even a half-step in that direction. They're nothing more than Steam for consoles. You download and install the game and it runs on your hardware. Doesn't require constant major pipeline, or major rendering farms on the server end.

It sounds like another example of people with a cool idea who have convinced themselves that's it's way better than it actually is and have no concept of real-world implementation and customer usage.

 

[The Purple Crusader]

Yah, i dont see this working out...

 

[CakeEater]

Kinda hoping it will work, but with a sense of fear, technology can't just leap like that, wait... that might be the network that evolves into Skynet!!

 

[Eldritch Warlord]

It sounds like another example of people with a cool idea who have convinced themselves that's it's way better than it actually is and have no concept of real-world implementation and customer usage.

Exactly what I was going to say.

A system like this cannot work on a reasonably large scale for years. And when it does happen I think it will be either EA or Microsoft, maybe Ubisoft, starting it.

 

[renegade826]

At least they are being different. The whole games industry is becoming a big epic fail!

All I can say it, thank good for PSN and XBLA

 

[roblikestoskate]

AHAHAHAHAHAHAHA

Thin clients suck even when you just run a standard productivity app on a local server, a videogame over the internet is going to suck so hard you can use it to clean your room.

Thin clients ALWAYS come with promises like "you'll never have to upgrade your hardware again!" and then users complain about the horrible performance compared to a conventional thick client and in the end the thin client usually gets abolished. I have no idea how often that scenario has been dealt with in the BOfH stories but I've seen it happen in real life too.

Hell, how are they going to fix the lag issues? Rewrite the laws of physics? Sorry but signals only propagate so fast. Add the overhead from all the hardware in between and the theoretical minimum latency isn't going to be pretty, never mind the real life situations.

Oh and don't forget about ISPs deciding they don't like how much bandwidth you're wasting...

This may be the only intelligent response in this entire thread. But to further my understanding, what makes thin client software run so poorly? Are there any creative solutions?

At least they are being different. The whole games industry is becoming a big epic fail!

All I can say it, thank good for PSN and XBLA

What English is that?

 

[Escapefromwhatever]

Thirty years after OnLive changes the gaming world as we know it:

"New Console Lets YOU Own the Games!"

A radical new console, being made by Nintendo, which dropped out of the console business after its Super Wii couldn't handle games being distributed over the internet (though it did have some nice virtual reality elements, as well as the standard Nintendo mainstays now seen on consoles such as Microsoft's Zbox), is being developed in order to let the player have his or her own personal library of games, rather than having them loaded up overnight on the innovative but slow OnLive service. This new console, the Nintendo Entertainment Console, or NEC, will use cheaply made flash-based cartridges to hold games, capable of holding more data than the Blu-ray discs of old. Nintendo recognizes the problems of developing a console based around the physical medium these days, but insists that gamers will go for it. They tout its benefits as being:

- Perfect for gamers with no online connection (Kids?)
- Having virtually no loading times
- No system crashes that will delete your data
- Installing a game to the NEC is as good as playing it over OnLive

Call me a fool, but I think this new-fangled approach is just what we need right now. The only competitor to OnLive right now, Microsoft, has unfortunately failed with its homegrown attempt to mimic the OnLive service. The only things keeping its gaming department alive right now are the popular Halo and Gears of War series, whose makers are looking to drop Microsoft as a publisher. Because of this, the only practical console (or program for the PC users) is OnLive, which has been known to be cold and businesslike, responding slowly to gamers who are plagued by lag or lose save data constantly.

Perhaps Nintendo's entrance into the race, which also promises to bring a fancy neuron controller to the table, as well as a classic controller for those who don't wish to use the new technology, will give OnLive much needed competition, forcing them to consider those who would rather figure out new hardware than deal with complicated and sometimes shaky software.

 

[philosophicalbastard]

Thirty years after OnLive changes the gaming world as we know it:

"New Console Lets YOU Own the Games!"

A radical new console, being made by Nintendo, which dropped out of the console business after its Super Wii couldn't handle games being distributed over the internet (though it did have some nice virtual reality elements, as well as the standard Nintendo mainstays now seen on consoles such as Microsoft's Zbox), is being developed in order to let the player have his or her own personal library of games, rather than having them loaded up overnight on the innovative but slow OnLive service. This new console, the Nintendo Entertainment Console, or NEC, will use cheaply made flash-based cartridges to hold games, capable of holding more data than the Blu-ray discs of old. Nintendo recognizes the problems of developing a console based around the physical medium these days, but insists that gamers will go for it. They tout its benefits as being:

- Perfect for gamers with no online connection (Kids?)
- Having virtually no loading times
- No system crashes that will delete your data
- Installing a game to the NEC is as good as playing it over OnLive

Call me a fool, but I think this new-fangled approach is just what we need right now. The only competitor to OnLive right now, Microsoft, has unfortunately failed with its homegrown attempt to mimic the OnLive service. The only things keeping its gaming department alive right now are the popular Halo and Gears of War series, whose makers are looking to drop Microsoft as a publisher. Because of this, the only practical console (or program for the PC users) is OnLive, which has been known to be cold and businesslike, responding slowly to gamers who are plagued by lag or lose save data constantly.

Perhaps Nintendo's entrance into the race, which also promises to bring a fancy neuron controller to the table, as well as a classic controller for those who don't wish to use the new technology, will give OnLive much needed competition, forcing them to consider those who would rather figure out new hardware than deal with complicated and sometimes shaky software.

I wouldn't be suprised if that actually happened.

 

[Blaxton]

I use a thin client on my phone that is capable of rending Hulu, Youtube, and almost all other streaming video. Without a server application doing the dirty work that would be nearly impossible on a standard windows mobile device. I won't lie, the browser has its fair share of problems, and it only runs at QVGA (which looks weak on my VGA screen), but it still does something that all the other browsers (5) that I have my phone can't, which is render pages like a desktop to 100% accuracy, plug-ins and all.

I'm not going to say that this technology is going to be the unbridled revolution that I so want it to be (spending over 3K on a rig to play a game on the highest settings a year after it comes out is just insane to me), but I think it's where we're headed. With 'netbook' as the new buzzword, you can be sure this type of software/hardware/user paradigm is going to become a big deal.

Latency is the only issue I can see being a problem here. I frequently use my home computer to connect to my work computer with a free remote access app. The video quality isn't great but the mouse clicks work well enough, in real time, to make the experience pretty seamless. Games, of course, are more demanding, and so it's going to test the capabilities of the concept to it's fullest.

I see this as being a proof of concept product, if it works only 50% of the capacity they are claiming, it's going to show that this is at least possible, and very probably the future of computer software.

 

[KDR_11k]

Generally thin clients have trouble because centralized processing power is usually more expensive than distributed power because hardware power and cost don't scale linearly, you can't get a central system that has 20x the power of a single system for the price of 20 single systems. Of course the solution there is to build a cluster so you essentially have 20 single systems stored in a central location networked so the clients don't see the different systems. Would probably work well enough in a productivity environment as your systems could still serve multiple users each but games tend to use 100% of the system (especially the not so easily scalable graphics memory) so you pretty much have one system per client. Sure, average user loads would probably mean you could oversell your capacities but then you get in trouble on peaks. Unlike an ISP you can't just allocate fewer resources to a user because he'll be impacted badly by that instead of merely having to dial back on a bit of luxury (1MBit is still good enough for most tasks even if you were promised 8, you can probably manage and most won't even notice the degradation). And that's just the computing troubles. The bandwidth won't be easy either, running 5MBit per user constantly isn't going to be cheap and again you get in trouble if you oversell and endup with more users wanting in than you can support. People probably won't like it if the service has a waiting queue. Then you get into latency. When you play online you see way less latency than there is, the game has tons of predictive algorithms that hide the lag (it'll at least let you look and move around without lag even if the server will sometimes pull you back a bit). A thin client can't do prediction so you'll feel the full strength of the lag. This is worse than usual even because it's I/O lag, not just lag between your simulation and the other players'. Means if you look around with the mouse you're going to see the reaction with delay. Worst case would be something like Guitar Hero or Street Fighter.

Lag isn't something you can get rid of with "network accelerator cards" or fancy algortihms (at least not with a thin client, you can of course hide it with a thick client), lag is a fundamental property of the physical connection between your system and the server. The physical signal only propagates at a certain speed, router and repeater hardware and such will inevitably add more delay to the signal. One frame in a 60fps game takes 16ms, on a 50ms lag you'd have a delay of about 3 frames on everything and AFAIK lag can easily go higher even in good conditions. And that's before we look at the feasability of a sustained 5MBit stream (with no buffering!) to a consumer connection.

 

[phirewind]

I use a thin client on my phone that is capable of rending Hulu, Youtube, and almost all other streaming video.

Streaming video means reading a file off a hard drive and sending it over the internet. Streaming gameplay means having a beast of a machine for each user actually running the game in memory (Crysis likes to have at least 1GB all by itself) and rendering each frame, then compressing it, THEN sending it over the internet. Most streaming video servers in place now that can handle hundreds of simultaneous live video feeds couldn't run ONE instance of Crysis at 1280x720p at 30fps.