After yesterday's
announcement, we've all wanted a first-hand look at OnLive's gaming performance. Can Steve Perlman, the creator of QuickTime and WebTV, really "change the landscape of the Games industry" using OnLive's new interactive video compression algorithm? Latency through the algorithm is just 1-ms instead of the 0.5- to 0.75-second lag inherent in conventional compression algorithms used in corporate video conferencing solutions, for example. While OnLive's backend servers do the heavy crunching, pretty much any PC or Mac has the power to decompress the video at what's perceived to be real-time. As Steve puts it, "video is trivial for us now." The demo starts at about 10 minutes into the video and looks damn impressive running on a
Dell Studio 15 (16 minutes in) -- yes, it runs Crysis. Controlled yes, but very, very promising. See for yourself in the video after the break.
P.S. Don't forget to
sign up for the Beta starting this summer.
Update: During the Q&A, the very real concern of network latency was raised (35:20 into the video). Perlman said that OnLive will work with a data center that is about 1,000 miles away from DSL or cable connections (probably 1,500 miles away with fiber). OnLive data centers already exist on the left and right coasts in the US with a 3rd coming on-line in the mid-west. When streamed to gamers in Australia "just for fun," OnLive found that "you can kind of play the game, but you see the lag."
posted March 25th 2009 5:31AM
...more like OnMyLaptop and classes be damned!
Doh!! signup page bad in google chrome, please don't make me use Firefox! :( .. too late, zombies have overtaken my pc once again ...
I would really like this for my crappy PC..
I still not 100% on this.
Can you use xbox live services on this?
Do you still need the wii bar and how do you plug in the controllers?
Can you use the ps3 blue tooth controllers on your pc thru bluetooth?
What about xbox live headsets?
All these things he didn't address
and if they don't let them will it be possible to hack them on to it?
Holy cow... this thing plays Crysis... If it runs Doom and blends at the same time I'll take two..
where the fuck is bungie and ensemble studio's that's like half the industry, I hope they will join sometime, halo3 on my netbook, ah it makes me tingle inside
@stephen
not now, lets the adults talk.
I think Video Card Companies, Publishers, Nintendo, Sony, and Xbox crapped themselves...
Its gonna be a good time to be a game developer
I call BULLSHAT. Have any of you ever encoded an HD video stream? Try doing it in realtime on a 60fps source. Its not gonna happen.
@junk: but isn't that supposed to be the breakthrough that onLive is? that they were able to find a way to encode video in real time and stream it for their users?
Anyway.. with my crappy DSL I'll only be able to play SD games when/if this comes.. but I don't mind.. it's still better than not being able to play anything ^^
So Phantom Gaming Console.... for the PC/Mac? Fantastic... FAIL
Other than the lack of actually modding these games, this seems pretty damn impressive. I might just get it instead of my next Rig Update that's due in a year or so. Man micro console is seriously micro too.
I guess that means Pandora CAN play Crysis...
If it has an Ethernet connection.... then theoretically it can play anything you want....
BUT:
Why would you do that? It won't work on the go (who has a mobile 2MB connection), so this is for home gaming. You would be wasting your money playing on a small device at home, even if you "can".
Maybe in 5 years when 3G advances... A LOT... along with battery technology.
in 5 months yeh, damn is this going to be like the Beatles to music as bungie to gaming? cause that would suck major balls
@ TareX:
In 5 years I' hope you are not using 3G, since most carriers will have switched to 4G (Yes even MetroPCS), and speeds will be similar to broadband, if not faster. Granted you will probably blow through your data limit by playing this game, but it will be possible.
In Sweden we already have turbo 3G with speeds up to 7mbit/s in the big cities. That is absolutly enough to use OnLive.
@Tarex
Presumably a smaller device, running a game at a lower resolution, would have smaller bandwidth needs. Of course, whatever handheld device you're using must have enough local horsepower to decompress the stream.
Imagine all the possibilitys, forget netbooks... You'll probably be able to play CRYSIS on your phone sometime in the future thanks to this technology.. I think I'm gonna start inventing a GAMING UMPC dedicated for this device. Who wants to go halves?
that's fantastic!
but how many servers will they need to run this? every user will use intensively 1 graphic card, so i guess they'll have to build pretty big server farms....
good question since graphics cards don't virtualize all that well, but who cares how many servers it takes. that's the point of the service model is that you let them worry about it. complaints will come in and they will have to react.
somehow, I don't think they'll have a 4850 for every user.
probably some uber future graphics card that the futuredudes sent back with the 4400
He said he was only 50 miles away from the server. If someone was say 600 miles away from the server, their lag would be much to great to play a game. So if you live within about 100 miles of that server, congratulations, your eligible for OnLive.
A well thought out comment. Most of the delays are at the consumer and server end. Not the middle. The extra distance you describe, 100miles to 600miles, would be principally traversed through the fibre network backbone of the internet. Light travels through fibre, light travels fast. 500 miles would take light 0.00267 seconds each way, roughly an added 0.005 seconds of latency. Are you a Hummingbird?
This is the future, the cloud is the future, the infrastructure will upgrade to the demands of the consumer and the market.
@Spanner
You dont have a direct cable from the server to you door. There are repeaters and routers along the way which require processing to route you packets so this certainly isn't the speed of light.
Spanner:
First, light travels significantly slower through a dense medium (like optical fiber) than through a theoretical vacuum. Second, there are multiple switching points between you and anywhere else on the Internet. I don't know about you, but I'm not wealthy enough to afford a direct fiber connection from my doorstep to major carrier hotels. Maybe some heavy peering would help, but you'd have to have *hard* guarantees of sub-10msec RTTs, at the least, to make this even remotely practical - and that's assuming their encoding technology is perfect, which I have my doubts about.
This is snake oil.
Regardless of how impressive their algorithm supposedly is, it's irrelevant and moot.
THIS SYSTEM RELIES ON AND IS BUILT TO WORK WITH EXISTING SYSTEMS -- existing systems that introduce noticable lag irrespective of any algorithm known to man.
It relies on gaming engines which take time to compute user input, it relies on the internet and unless we are all running fat uncapped pipes with virtually no latency and live within walking distance from the servers/mainframes there will be lag, it relies on client/server architecture and authentication systems which will introduce lag, and any system that works in conjunction with the pre-existing architecture and systems already in use will have noticable lag no matter what these snake oil salesmen say otherwise.
They may have predictive algorithms in place which will attempt to render scenes in anticipation of player movements, but that won't work for FPS's, fighting games, racing games, and a plethora of others.
I can see a system such as this being used for MMO's or Turn-Based Strategy/RPG's, but these guys are trying to con a lot of naive people out of some cash, and I for one can not believe that the readership here for the most part doesn't know any damned better.
I wonder why so many people are so negative about all this. I mean, sure, what you say is all true and I hardly believe that this is going to work flawlessly and without lag either. But nonetheless, I also think that they sure thought of that problem. And who knows, maybe they have some solutions in mind and an ace up their sleeves.. How many times achieved someone something that most people believed will never be possible? I'd say many times.
So, let's just not get to negative, wait and see what they gonna create. It is either gonna be revolutionary or it ain't. But at the end of the day, if it's gonna work: Great! Otherwise: Whatever..!
Did you even watch the video? He clearly states that they have made custom silicon and that the servers are ' unlike any other out there '. I mean, I'm pretty skeptical too (as well as hopeful) but at least I watched the video :P
What brand of tinfoil do you use?
Mark,
The type of tinfoil hat made out of computer engineering and information systems degrees which have taught me that the promises these charlatans are making lie outside the realms of physics, electrical and communication systems.
It's simply feasible for the masses.
Also the type of tinfoil hat made out of an MBA degree which has sharpened my senses and allows me to better detect con artists when I hear, see, or read about one.
And above all, the type of tinfoil hat made out of common sense.
Like I said, I can't believe how many of you actually fall for their rhetoric so easily without any reservations.
Anyone can get an MBA nowadays. What actual qualifications do you have? Steve Perlamn has developed Quicktime and Moxi.
So, is that the heavy duty Reynolds wrap then?
So he's developed one (pardon the french) shitty format and a rather lackluster media device.
Great, so he's also rewriting the laws of physics and communication systems too is he?
Someone hand this man the nobel prize.
to be fair to Hamidxa - I too have reservations on the application of this in a commercial sense.
I believe yes it maybe technically possible - but commercially no unless you charge people more over the course of a year than what it would cost them for a console.
Its all well and good to show a demo where only 2 people are making use of the service but try and scale that to 1000, 10000, 100000 and this seems laughable.
This can only work by invitation only or "please wait x numer of days before using the service".
How many servers will it take for a 32 player online game?
Not too sure about this move back to mainframe computing.
Peace out
I still don't understand how they intend on reducing my already >16ms packet lag to 1ms.
Heck, I just did a ping from my laptop to Chicago (closest major city, over wifi that adds lag) and I'm averaging 40ms. You can't make that faster (besides plugging directly into my router and moving closer to a bigger city.) These are 64 byte pings. Any kind of precision control information will be up there (analog stick floating point numbers + button arrays + motion controls...)
--- chi.speakeasy.net ping statistics ---
28 packets transmitted, 27 received, 3% packet loss, time 27126ms
rtt min/avg/max/mdev = 35.965/40.998/76.740/7.962 ms
Exactly Andir and Poke, thank you for injecting a breath of fresh air and common sense into this "thread".
Technical limitations aside (as Andir and I pointed out), economically speaking this is equally infeasible and impractical.
Just how do they plan on providing not 1, not 2, not even a roomful of people (a "controlled" roomful at that), but a million, how about 10 million, or perhaps 50 million on a good day, enough computing power and bandwidth (on their end) to satisfy the needs of HD gaming?
It's simply not feasible, not with the current internet infrastructure and pricing models.
If this model does indeed one day supersede, if not altogether replace, the current one, then the infrastructure needs to be in place to faciliate that sort of transition.
However, such is not the case now nor in the forseeable future, at least not for several more years.
Open your eyes people, this is NOT ready for the masses.
If we're all still alive when this thing launches later this year, I'll be back to say I told you so.
Maybe they are aiming for "good enough" latency rather than competitive gaming latency. Fact of the matter is there are millions of gamers out there who can't tell the difference between a 40 and a 100ms gaming experience. Hell I used to play Quake1 over dialup with a 320ms ping and loved every minute of it.
I will reserve judgment on this until it's properly reviewed. I don't see the point in taking a dump on this idea like some people here insist on doing when they have never actually used it and they are basing their opinions on pure conjecture.
@ChuChu you would notice the latency on this... conventional networked games are rendered LOCALLY with next to no latency .... the server only coordinates where you see the other players ... basically conventionally a latency hiding scheme occurs which would be completely useless with this method
say you have a 100ms ping thats gonna be 100ms + a few ms for the game logic to respond + 1ms to encode as i see it + 1 ms to decode
they might be able to eliminate some of the last three but that big 100ms chunk ain't goin' nowhere anytime soo unless they figure out a way to pass the internet through a time warp sub space thingy
There's a bit of a difference when you're only sending move information. The client, if set up properly, can estimate your location based on your former report. This is sending compressed video to the client. When you hit a button on my laptop, it will take at best 100ms for the signal to get to the server, render a frame, and return that image to me. People already complain about rhythm games like Guitar Hero being impossible to play on TVs with as low as 30ms input delays. I can't imagine someone trying to play this over a laggy Internet connection on top of a TV set that has a delay factor built in. Some TVs delay signals due to scaling as well. If you aren't using a native resolution, it could be even worse! Quake over a 320ms modem is at least responsive to the player even if the other player tends to warp every now and then.
@ Hamidxa
Why did you put 'thread' in quotes? Are you implying this is not a thread?
this is gonna cost more than your mom in hot lingerie
What if you're mom is a hooger?
Someone who lives in Hooge? http://nl.wikipedia.org/wiki/Hooge
Hooge looks like a nice place.
This could bring gaming to Linux if done right. Although i'm sure nvidia wouldn't be too fond of this idea.
What does nvidia have to do with it? It's purely software.
Actually nVidia is gonna love it. More of its stuff sold in a recession.
Well would you buy a 280 to just stream media? Nope you are going to opt for the low end scale to save cash. I could see this very well hurting their bottom line.
Except that these servers will need some serious graphics hardware to run multiple instances of these games, and that hardware has to come from somewhere. What did you think was going to handle rendering, an array of hamster wheels?