OnLive MicroConsole to be Free?

Slashgamer Writes: There has hardly been any news since the first reveal of OnLive to the world at GDC 2009 in San Francisco. When Steve Perlman, the CEO of the games-on-demand service and one of the inventors of Apple's QuickTime technology, took center stage at the press briefing to introduce the gaming community to his newest project OnLive - he was met with both praise and criticism. Some people thought this ambitious "cloud-gaming" service was the future of gaming, while many others thought there is no way this technology will work with the functionality of today's Internet capabilities. But with all these questions, most people always seem to ask the same one; how much will the consumer have to spend in order to use the service?

A few days ago, a video of Perlman was uploaded onto youtube, where he went back to his old school Columbia University and discussed his next endeavor. While nothing to new was unveiled, he did briefly mentioned something interesting about the MicroConsole."

[NotSoSilentBob]

I'm thinking it will be like a cable box, they require you to have it for their "broadcasts" and then charge you a monthly renters fee.

[Proxy]

But I doubt the industry will change much. If OnLive charges per the amount of time you spend playing, then some game are going to need to charge around $20 per hour to match their current profits. :)

In other words, if devs want big money, they'll have to deliver something to keep your attention for hours and hours. (Even more difficult in the OnLive environment due to the option of "just trying" so many different games. You could spend over half your time playing a level here, a level there.

[Golfcoachh]

I mean come on, I have a 16mb per second cable connection, and I still have a hard time getting downloads faster than 1 to 3 mb per second. I can just imagine how much computing power they must have to have if htey are going to have millions of people playing at the same time, possibley the same game. I also hope this isn't the end of the single player campaigns in games. I want it to be awesome, but with most of the U.S. at least struggling to get more than 2mb per second service, I thing this is still a few years away. Not to metnion companies like comcast capping downloads and then charging a premium once you go over the ammount. So how many gb of info will be streamed in 8 hours of gaming in a day?

[Bigpappy]

They need some serious server support to run full games remotely, with millions connected to the same game. There is no off line component. You have to pay monthly fees. I believe digital downloaded games is the real future of gaming. Not cloud or stream gaming. Having the game local will always be the preferred and fastest way to game.

[SilentNegotiator]

...with a year subscription.

I'm not interested either way. It's like taking PC gaming, taking away all the good things that it has over consoles, give it even worse frames-per-second and worse image quality than on consoles, and be completely reliant on a constant internet connection. And once they stop getting your money or die off, you're left with a plastic box. A useless plastic box.

[DeadlyFire]

Why does everyone worry. Either way OnLive, OTOY, GaiKai and others are a big plus for the mini PC and Netbook market with cheap PCs that don't play games. Now they can through stuff like this. PC gaming as a whole will still be PC gaming. Only difference is that OnLive will be another platform inside of it. Not much really changes.

I think OnLive, OTOY will both be great with their services. I can't wait to learn more at GDC and E3 2010.

[AliTheBrit19]

It could be more like a monthly service..that would be better

Either way I see OnLive being very expensive

[penguinhunter]

I'm not sure if OnLive will fail, but either way I'm interested in what it can do.

[KageMonkey]

OnLive is snake oil for most games. They claim a 100- ms round-trip latency from button push to display, which is highly optimistic.

==========
They claim:
==========

Player Input Sent to Server (25ms - Highly Optimistic)
|
Server Process Input, Generate Image, Compression (5ms - Optimistic)
|
Send 720p Image (200kB, no Keyframing @ 10Mbps = 20ms - Highly Optimistic)
|
Data Propagation (25ms - Highly Optimistic)
|
Image Decompression (10ms - Likely)

==========
Real World:
==========

Player Input Sent to Server (50+ms)
|
Server Process Input, Generate Image, Compression (5+ms)
|
Send 720p Image (200+kB, no Keyframing: 30ms)
|
Data Propagation (50+ms)
|
Image Decompression (10ms)

In the real world, even assuming ideal situations with no packet loss and no backlog of data, we are looking at 100-200ms of latency.

Basically, it is similar to doubling the latency on all the online games that we play today, and an additional sluggishness produced by packet loss, network traffic, and numerous other networking factors as the game is no longer client-side.

It might work for games like LittleBigPlanet, but it will not work on FPS and Fighting Games.

[masterg]

@KageMonkey

Did you not see the 48 minute presentation where he showed how all the problems you talk about were solved? They have it running with a 80ms round trip.

Millions of gamers is not a problem either.

They rent high scale serves that each can handle a number of players. A server filled with players makes profit. Meaning every time you start running out of servers you just rent another one. More server = More profit.

After seeing the presentation video a few days ago I have been convinced that this service is the next big thing. If I were Sony of MS I would be worried and start thinking about buying this company so they could offer the service themselves. Just imagine a PS3 with Onlive. That would mean that they had 99% of the games coming out in next gen.

[KageMonkey]

Yes, I saw the presentation. What they provided in the presentation is an overly optimistic view of real-world application.

Even with proper data center placement, round-trips between the server and client are NOT guaranteed to be under 50ms.

25ms latency is almost ideal for online gaming. However, such latency is only achievable at lower transmission rates with ideal data center placement.

When significantly more data is going through the connection, it is theoretically impossible to achieve such latency due to packet loss and switching. You can't just "solve" network architecture and claim a 80ms round-trip.

Like I have illustrated in my first post, real-world round-trip for high throughput applications are closer to 100-200ms.

[bozebo]

exactly KageMonkey.
for some reason the world is full of idiots who don't understand.

whoever disagreed with you is an ignorant fool.
even with fibre connections to all homes, the lag will be very noticable. not only that but most people have inferior routers, and bad contention ratios. so the connection will be inconsistant. even if something like 50ms was achieved there would still be noticable input lag, because the packets wont necessarily be aligned with the frame rate - then take into account tv input lag. any competent gamer would notice the lag: imagine you need to aim for somebody in an fps game, but they move - meaning you need to react and aim in another direction; all this takes place subconciously, and that is when you rage out at the person surviving like 40 bullets :D - because client side you probably played perfectly. This increases the actual delay in reactions exponentially, though the illusion is that there is less lag than you think - with OnLive this effect will be greatly exaggerated because the picture is streamed from the server itself.

[TheIneffableBob]

The OnLive TV device only costs around $25-30 to manufacture. Very cheap.

It's going to work the same way telecoms give cellphones for free. They give you the device but you'll have to use the service, and they make significantly more from you using the service.

[HDgamer]

The service may cost an arm and a leg. I'll keep up with onlive but I doubt i'll use it if it's not what they claim it is.

[drewboy704]

I dont see onlive making a move in the console industry although I think steam should be worried since onlive should work well on the PC.

[bozebo]

itd better be free with the epic input lag that will occur. love the way they lie and say there will be no lag, lol what idiots - clearly don't understand networking.

[Crazyglues]

http://www.gamertagradio.co...

Let's not forget this is the guy who invented Quicktime, so if he says they found a way to get no lag, I would say changes are he did..

And after watch the whole presentation I was really impressed, this guy is really, really smart and they actually have made something amazing here, that will change gaming forever... and the way we play games..

It really is going to be an Amazing service.. I can't wait to play some Crysis.. LOL (just to see how good their high end cloud servers really are)

[masterg]

@Crazyglues

I agree with you 100%.
I saw the video and was blown away.
He got really technical. All the problems they had faced would have send a normal company running.

People need to understand how this works.
Normally when you play online you have people from all over the country/world playing. Here you will only have people who are within a certain range.

Normal when you live in LA and play a server in Dallas, you might take a route than goes through Seattle --> Denver --> Pittsburgh before you reach your destination.
Here you will not play a server in Dallas if you live in LA. You might play a server in LA or San Fransisco. And your route will go there directly and back.

[bozebo]

you 2 are ignorant fools, exactly the idiots that marketing intends to capture. You clearly know NOTHING.

Yes, it is possible to have people nearby playing together, networks like xbox live already try this. But what if you live in an out of the way area? You are forced to play with more distant people. But that is besides the point entirely :-
It only matters where the servers are, this is not a p2p system. The routing, or hops in this case, is NOT under OnLive's control in any way. Where the packets go is determined by the ISP. Not only that, but even if there is ONE hop, the lag would be FAR too much for it to work. Anything more than perhaps a consistent 30ms would be extremely noticable, as it would take longer than one frame for your input to get back to you.