By Jim Rossignol on March 24th, 2009 at 12:19 pm.
At the last GDC the industry big brains were sat around telling us how games would one day be remotely rendered on big computing clusters and then streamed to our TVs. The big unveil at this year’s GDC has proved them to be correct. Maybe. OnLive is a service on which you use superfast broadband (1.5mbps minimum) to play games on a remote server. You just plug it in to any “entry level” PC or Mac, or hook it up to your TV, and play. It doesn’t matter if you don’t have the latest 3D card: because the remote server does the rendering and streams the result to you. That’s the theory anyway, and it’s a theory a bunch of big name publishers have signed up to. Watch the OnLive spokesman Steve Perlman make his big claims after the jump.
24/03/2009 at 12:21 teo says:
Pretty sure he says mbit
also, won’t work
24/03/2009 at 12:21 schizoslayer says:
I don’t think this will work for the same reason DRM on Music didn’t work. What happens when the servers shut down?
24/03/2009 at 12:26 The Sombrero Kid says:
I’ve said it before and I’ll say it again, never going to happen people where claiming this has a round trip latency of less than 1ms which is fucking impossible cause i can’t even ping my isp in less than a millisecond absolute vapourware guaranteed, also i and I’m sure the rest of the internet am unwilling to give up my unofficial ownership of games unless it was a lot cheaper (read virtually free).
24/03/2009 at 12:28 schizoslayer says:
I’ve just realised that my argument is flawed as Steam has the same problem. However at least with Steam I can play my games when I don’t have an internet connection as they exist on my PC and not in a cloud somewhere.
24/03/2009 at 12:30 Yargh says:
The idea of leasing some games and having all of your saves available from wherever you wish to play, not to mention being able to avoid the never ending PC upgrade cycle has it’s attractions.
On the downside you’d better have a rock solid connection on an unlimited (really unlimited, no fair use rules) use contract to have a chance at it working for you. Not to mention that Euro-players don’t have chance until they run a server farm over here.
I kind of doubt the current ISPs are going to be too happy about this kind of thing, they’re going to want to sell us ‘gamer quality’ connections at a premium as their existing price models are unsustainable if many people actually use what they pay for.
24/03/2009 at 12:31 AbyssUK says:
Good idea and one day this will be how it all works. But not yet.. maybe 10-15 years from now.
24/03/2009 at 12:31 Matt says:
@Sombrero
He sais the video compression algorithim performed in less than 1ms, he doesn’t claim the round time delay is 1ms.
IGN who actually checked it out say they’re perfectly playable: http://uk.pc.ign.com/articles/965/965535p2.html
TBH I think this is the most exciting development of cloud computing I’ve seen.
24/03/2009 at 12:31 Robin says:
Won’t work.
Don’t be fooled by the list of publisher partners. Just means they’ve been offered free money/marketing.
24/03/2009 at 12:34 kyrieee says:
Sombrero 1 ms haha that bullsh*t
It takes 30 ms for light to travel from the US east coast to the west coast and back, and packages are slower than that. Of course, you might be closer, but in 1 ms light travels 300 km, so 150 km one way, and they can’t possibly have server farms that close to everyone
Also, 5 mbit/s isn’t enough. I mean, that’s 600kb/s which is 10kb every 1/60th of a second (each frame in the game). That’s NOTHING for an image that large. Video works with a 600kb/s connection because it’s pre-rendered, which takes a ton of CPU power and can’t be done in real time.
Pipe dream
24/03/2009 at 12:35 framstick says:
has anyone actually used this in a ‘real world’ situation…ie one not controlled by them? …any beta testers?
24/03/2009 at 12:36 bansama says:
This is a horrible idea and if it’s the future of gaming, then my gaming days are numbered. Aren’t we already too reliant on the net? An infrastructure that’s just going to buckle under moves such as this. (Look at the backlash with ISPs in the UK over the iPLayer or whatever the beeb called it).
And what happens when there is a crash? No local way of backing up your gaming progress? What if there’s an issue with your ‘net connection? How about those on metered connections who already have a hard enough trouble with Steam and the like?
Nope, perhaps I’m just old fashioned, but I really have a horrible feeling with this cloud computing stuff. I just have a nagging feeling that it’s not going to work even if I can’t exactly pin point why. Give me data stored locally on my own PC any day. Data that doesn’t rely 100% on a service of which I have no direct control.
24/03/2009 at 12:37 The Sombrero Kid says:
@Matt
woops good catch, well my point about this last year stands this paridgram is about where people want thier computing power located in thier house or in a server farm i would say both, people aren’t going to rely completely on this service mainly because it’s cheaper to run your own pc, there’s nothing stopping an individual from running the same setup at home and achieving all the benefits with none of the drawbacks i.e. paying for it or losing control of your games.
24/03/2009 at 12:44 Stijn says:
Doesn’t this already exist? I remember submitting news about a similar concept to some big local tech site about 2 years ago.
Ah, http://www.streammygame.com/smg/index.php
24/03/2009 at 12:46 Matt says:
Well the benifits are obvious. Let’s say you’re going to spend £40 on crysis. If you want to play it on your own pc at high settings then you’re going to need a computer that has an expensive graphics card and a powerful cpu which can set you back 200-300.
Or you could buy it on this service, and be able to play it on a netbook that costs as little as £150.
I’m not saying that this will replace traditional games, people sitll enjoy playing resolutions of 1920×1200 (i.e. me :D) but for netbooks, people who can’t afford expensive computers etc, this is an incredible idea.
24/03/2009 at 12:46 Tworak says:
input lag… yeah… no thanks. it might be good if my ISP hosted the thing and I’d get is gonna suuuuuuuuuuuuuuck.
24/03/2009 at 12:48 Tworak says:
^ got my words eaten!
it might be good if my ISP hosted the thing and I’d get 5ms but 50ms+ is gonna suuuuuuuuuuuuuuck.
24/03/2009 at 12:50 Matt says:
@Stijn
That service is designed for lan situations where the bandwidth is 100mb
24/03/2009 at 12:52 Daniel Puzey says:
Bandwidth nonwithstanding, imagine 100,000 users playing at once. Have they really got 100,000 GeForce 280s sat at the other end running all this? If not, how do they render it all for all those users at once?
It’s not like you can just cluster a bunch of GPUs together and split the work up (as you could if this were regular CPU-bound work). So 6 million frames per second… how?
(And I’d presume they’re aiming for more than 100k peak users, too.)
24/03/2009 at 12:55 kyrieee says:
Giantbomb says the latency is 80 ms, which is terrible
24/03/2009 at 12:56 redrain85 says:
Yeah, of course all the big game publishers are slobbering over this. The games as a service model in its purest form. The customer no longer owns the game. It becomes a permanent rental. An evergreen source of income. Or, so they hope.
It’s interesting to postulate, however, that a service like this could mean the death of console gaming. Yeah, bet you never thought you’d hear that phrase. *cackles*
For all the people who complain about how often, and expensive, it is to upgrade their PC: they could buy a dirt cheap PC and still play the games.
And as long as publishers do still offer their games for sale at retail – with this streaming service remaining an option, for the people who really want it – what’s running on the render cloud can run locally on your PC, with the right hardware. Since the render clouds just happen to be PCs, too.
This could potentially be a great boon to the development of PC games, depending on exactly how things unfold.
24/03/2009 at 12:56 framstick says:
actually- thinking about it we use a system called logmein for work – i have used it all over the world- it works well for remote control of a pc….if they can improve the video streaming – job done- it is free to try…..it basically only uses about 70k a sec so if you used more bandwidth you could update picture more often…i think it its doable…
24/03/2009 at 12:59 Malcolm says:
Hmm. Anyone with experience of using Remote desktop over the internet will probably be as sceptical as me about this.
Online games (Team Fortress 2, Unreal Tournament, Quake etc) do a lot of work to compensate for input lag – how on earth are they planning to retrofit that?
24/03/2009 at 13:00 Azhrarn says:
Sombrero 1 ms haha that bullsh*t
It takes 30 ms for light to travel from the US east coast to the west coast and back, and packages are slower than that. Of course, you might be closer, but in 1 ms light travels 300 km, so 150 km one way, and they can’t possibly have server farms that close to everyone
Also, 5 mbit/s isn’t enough. I mean, that’s 600kb/s which is 10kb every 1/60th of a second (each frame in the game). That’s NOTHING for an image that large. Video works with a 600kb/s connection because it’s pre-rendered, which takes a ton of CPU power and can’t be done in real time.
Pipe dream
With perfect routing, it’s technically possible to get really low latency over fairly long distances, however most countries don’t have anything even approaching perfect routing. The US and UK for instance both have aging infrastructure which really can’t provide fast links like that.
Many smaller european countries have more modern digital infrastructures, but even those are far from perfect.
1ms processing time server-side is possible, travel time will ofcourse be longer even with perfect routing, they can’t have servers within less than 1000 kilometres of every computer of the planet, now can they. ^^
24/03/2009 at 13:01 JJ says:
I remember RPS reporting on a MMO working like this. Sometime last summer IIRC. Cant find it at the moment
24/03/2009 at 13:02 Daniel Puzey says:
Remote Desktop doesn’t really count, because there’s an amount of hardware virtualisation going on there – if you ask a machine you’re remoted to whether it has a DirectX GPU, the answer will be “no”.
That said, playing games over VNC is a closer example, and not one that’s any more enjoyable!
24/03/2009 at 13:05 Sp4rkR4t says:
This type of crap has been around for years and will never work even when we all have fibre to the home because of latency.
24/03/2009 at 13:07 Matt says:
Ok @ all the people worried about lag.
A.) When is lag discernable?
Take CSS for example, an average ping of 40ms on your scoreboard means that the round trip time is 80ms + however long it takes the server to process the info but we’ll assume it’s 80 as that matches the giantbomb article.
Do you notice delay in css? Not really, I play CSS fairly competitively and 80ms delay is inperceviable in terms of when you expect a hit to register and so on.
Apply this to UT3 as malcolm suggested. UT3 unlike CSS doesn’t actually use client side lag compensation (this being, the client predicting what it thinks will happen vs. what the server decides happens (i.e. css) ) In ut3 a delay of 40ms with no lag compensation is again undetectable.
Using the remote desktop model. I remote in to my uni’s lab computers frequently, and there is no percivable delay from pressing a key on my keyboard and the letter appearing on the screen at the other end, and then being routed back to me. Which suggests that the delay won’t be noticable using this system either.
@Daniel
Worst case scenario lets assume it’s 1 computer for 1 game for 1 person. A nieve assumption that the computer costs £600 and people buy 10 titles a year means that the single computer will turn a profit 1/2 way through the second year.
If 1 computer can turn a profit then it’s just a case of the scalable economics.
24/03/2009 at 13:09 Rich_P says:
In THE FUTURE, when PCs have 50+ cores and applications are actually programmed to use more than one of them, I predict people subscribing to videogame services. So you could run Steam or a virtual Playstation 5 or Xbox 1800 and stream the content to your display of choice, using whatever input device best suits the game. So the actual console might live inside my bedroom PC, but I could access the service through my TV. If I wanted to play with a mouse and keyboard game, I just sit down at the PC and access whatever service provides those games, assuming people are still using KB+M for basic computing tasks.
Either that, or we’ll all be plugged into the matrix by then.
24/03/2009 at 13:13 Markoff Chaney says:
I’m sure in a perfectly controlled environment with a server farm in the next room connected via fiber this is quite enjoyable. Any other confluence of distance and/or media will lead toward this system being unusable, at least from a gamer’s perspective who depends on split second reaction times. We use a Citrix based virtualized environment at work and it can barely handle a decent number of people working, much less if they were streaming HD video constantly while still detecting and relaying input. Again, for those right next to the server farm, they have no issues. For those of us half way across the country, it’s not enjoyable at all. I shudder to see the lag from this kind of set up.
Wet dream for devs/pubs? Absolutely. It’s the perfect licensing method where you don’t even get the 0s and 1s any more to keep/save/modify at will, only the output the code displays to your monitor. Can’t Manage your Rights Digitally much better than that. You have a right to the output, not the game.
I prefer our current method and I’ll keep a fat client as long as I’m alive to play games on, even if it’s just Peggle. Not that I don’t like the idea of other computers carrying some of the load, but I’d rather run the load myself and know who to blame when I get killed.
24/03/2009 at 13:17 Colthor says:
@Matt
Ping is round-trip time. Ie. a ping of 40ms means a round-trip of 40ms, not 80ms.
24/03/2009 at 13:17 Catastrophe says:
Ever played a game over a Serial connection? This will be slower than that. And ohmygod Serial is slow.
24/03/2009 at 13:20 Jetsetlemming says:
80ms lag is imperceptible for most people when it’s lag on reporting where you are in an online game. 80ms lag for your key inputs? Fuuuck, no way. And that 80ms was reported at GDC where presumably they’re test hosting the service right there, not halfway across the country from your house on your home DSL.
Not to mention putting online multiplayer on this service would just compound those issues indefinitely.
Overall, I’d say even the basic concept is useless to us gamers, really, the only advantage is for the shitless publishers who would agree to anything to keep total control over their property. It’d be cheaper and way higher quality in the long run to just shell out the $700 for a decent gaming PC and actually buy games than buy their little “microconsole” for whoever much they’re selling it for, subscribe to their service (which is probably going to cost a ridiculous amount a month to support all those big name publishers with their big name console games), and play them with bad control lag and worse update lag.
24/03/2009 at 13:22 Max says:
@Matt
You are talking about network lag, which is different from input lag, although through this service, these two become one.
The 80ms are the delay between moving your mouse, and the screen reflecting this movement. Thus the game will be as inresponsive as if you were playing it with 12.5 fps.
24/03/2009 at 13:22 weegosan says:
Do you notice delay in css? Not really, I play CSS fairly competitively and 80ms delay is inperceviable in terms of when you expect a hit to register and so on.
That’s because they use smoothing algorithms that modify the reality of the 1s and 0s as the server sees it to make it feel it is playing how we perceive it. That is result compensation. What everyone is talking about is action lag, the other side of the coin. If you’re actions are lagging behind then you get a major sense of disconnection from the game and it will be shitty to play. There is no smoothing algorithm for what the player might do in the future.
24/03/2009 at 13:25 DarthInsinuate says:
I’ve just invented a free-energy device. It’s powering this computer.
24/03/2009 at 13:26 jph wacheski says:
sounds like a ploy to get investor money into a vaporware project,. and it really is a dumb idea as the net is already too congested with all the hi-def video and p2p torrent usage,. who has edless bandwidth anyway,. and of course the latency would be wack,. unless your playing Myst or some sht,. .
24/03/2009 at 13:27 Rich_P says:
Even if this service fails, I think it’s indicative of things to come. 30 years from now, I simply can’t imagine people still buying videogame consoles or even PC upgrades like they do today. Subscribing to videogames like you would, say, cable TV seems more reasonable. Input devices would really be the only hardware required. Games would no longer be confined to a given system spec. Compatibility would be guaranteed.
24/03/2009 at 13:32 danielcardigan says:
Is this for PCs? Surely this is an application to let you play HD games on your HD TV, using your cable TV Set Top Box. Why on earth would you need a PC at all?
24/03/2009 at 13:34 Downloads_Plz says:
Didn’t Valve have a project some years back that generated a lot of interest because they claimed it would basically completely eliminate lag?
And then they realized they couldn’t actually do it?
OnLive will generate a ton of interest because, well, it’s a pretty damn interesting project. But at the end of the day, it will stay just a project, and quite likely never actually see release.
24/03/2009 at 13:37 The Sombrero Kid says:
@Matt
you are taking about lag between your world and the ‘server world’ syncing it’s not noticable because your world runs without the server world, what this is proposing would mean the jumping teleporting and false actions you see all the other players do in cs at 100ms would be happening with your own input too, so you turn the mouse and nothing happens for a quarter of a second and then you violently swing right and the guy your trying to shoot has run left, finally you get the crosshairs over him but he’s 200ms away from where he is on your screen (the distance between his pc and the server and the distance between your pc and the server combined) you hit fire but by the time your command gets to the server he’s just reported yet another movement making him a total of 300ms away from where you thought he was, compared with the client version where people are accurately dead reckoned most of the time you will actually end up with less that 100ms real world difference between his location and where you saw him on screen.
24/03/2009 at 13:38 Tei says:
This could work for really simple games, like point and click adventures and puzzles and porn.
But is tryiing to sell it to people that need 10 ms latency on the responsiveness of the controls, and games that are designed around that.
won’t work
Archive this on “try again in 10 years” departament, and make it for something else, maybe casuals and porn
24/03/2009 at 13:41 Tei says:
Quake1 was the latest game with pure client – server architecture. All other games, from QuakeWorlds *need* fat clients, that emulate phisics on the clientside to make *predictions*.
If you want to emulate “OnLive”, buy a wireless mouse that adds 300 ms to your input, and play with it Quake1 (netquake).
24/03/2009 at 13:41 skalpadda says:
40-100Mbit fibre is slowly but steadily becoming the standard here in Sweden if you live in a medium or larger city and as far as I’ve heard it’s going in that direction in other European countries as well. I doubt bandwidth will be a limiting factor, at least in terms of having a decent pool of potential customers.
I really can’t see this working on people’s wobbly DSL lines though, both in terms of latency and the actual throughput of the tech. This is not speaking as someone with in-depth tech knowledge but rather from experience of what you actually get from your broadband and latency when playing various games online. It just doesn’t seem feasible other than in a demo situation with a limited number of players.
And how on earth are they going to get enough processing power to run this at a decent cost? As others have said, 60 full quality images per second for thousands of users that then need to be compressed, sent over the intertubes and displayed on the user’s hardware. It seems like an awful waste of resources to me.
24/03/2009 at 13:43 Robin says:
The more I think about this, the stupider it sounds. It’s a classic example of a technical solution looking in vain for a commercial application.
If you own a good enough PC to stream HD video at 60fps, you can play plenty of games on it, and you can definitely afford an Xbox 360.
If you want to use this service for PC games, you lose nearly all the benefits of the PC as a games platform. No mods, no configurability, no playing offline, input lag, variable image quality and framerate.
This is a service built on a myth, that there are lots of PC games that require supercomputers to run, and a “gaming PC” is a huge investment, rather than £500-odd quid for something that will run anything which uses the 360/PS3 as the target platform for the next few years.
24/03/2009 at 13:43 Gap Gen says:
Well, for slower-paced games this could work fine. Probably won’t work for twitch-gaming FPSs or whatever.
24/03/2009 at 13:45 dsmart says:
The last time I saw/heard about something remotely like this, it was called the Phantom console.
Then it was DISCover or somesuch.
There are so many reasons why this is going to fail, I simply don’t have the will to right them all up.
If all they’re going to be doing is pushing adventure games, casual games etc – then maybe – maybe – they stand a snowball in hell’s chance. Any game that requires fast real-time response, is going to suffer as a result. Even if its on a dedicated FIOS network.
This looks to me like more investor money down the drain.
And listing publishers doesn’t mean squat. Phantom, DISCover and everyone else had those too.
In fact, so far only something like Zeebo actually has something that works, has games etc. If you live in a third world country – or Brazil – and want to play generations old (Quake anyone?) games.
24/03/2009 at 13:47 Heliocentric says:
This technology is already in place for browsing, the opera mini browser performs the rendering and then sends a simplified version to your phone.
So a phone can see still versions of any site no matter how media rich. If the mobility of a device ever causes it to be underpowered you are more likely looking at mobile phoes then laptops. The new laptop cpus are pretty bloody beefy.
24/03/2009 at 13:56 dsmart says:
Yah, go ahead, compare the rendering of a web page to that of a real-time game. I dare you! :)
24/03/2009 at 13:56 Stijn says:
Matt: Ah I see, it changed. It used to be pretty much like this OnLive thing in concept. Guess they didn’t change it for nothing…
24/03/2009 at 13:57 Skurmedel says:
Don’t really care if it’s possible or not. I’m concerned about the ownership issue. I use Steam alot, and already I’m concerned about the fact that I can’t resell my games. Not that I would do if I could, but that’s not the issue.
24/03/2009 at 14:00 Heliocentric says:
real time games? okay maybe I’m pushing it, but still its a useful application rather than a useless function which cost outstrips its form.
example, steaming video on a mobile works, that stream could be a game.
24/03/2009 at 14:01 Matt says:
@Max yeah you’re right, the 12.5 fps analogy is actually probably the best way to put it.
However, I’m still not convinced that it’ll be that noticable. Would be interesting to try out a little app that articifically delays all mouse+keyboard input for 80ms to see what it’s like.
24/03/2009 at 14:01 flo says:
skalpadda: yeah we all know the scandinavian countries are down with broadband speeds :) However I think “it’s going in that direction in other European countries” is being way too optimistic. I’m in germany not directly in a big city, but close to several bigger cities, and the fastest I can possibly get is 32 Mbit/s through cable, that’s the highest speed offered by the cable owner. Maybe directly in the center of a big city you could get a business connection, with much better speed, but last time I checked prices were insane.
24/03/2009 at 14:05 Pani says:
Aside from all the technical difficulties already mentioned and the fact that you never have a copy of it etc, there are some nice ideas put forward.
The community aspect sounds interesting, as well as being able to spectate on anyone and any game no matter if you own it or not. I like that idea. What a great way to get excited about games. Also, its like turning on the tele. Hmm, what shall I watch today? Red Alert, Left 4 Dead or a WoW raid?
24/03/2009 at 14:06 Irish Al says:
Shifting 30fps to say a concurrent 50,000 users? Good luck with that.
24/03/2009 at 14:07 Gap Gen says:
Steve Perlman looks perpetually terrified.
24/03/2009 at 14:07 Cedge says:
This is fucking ridiculously stupid. It’s not going to work. The video quality will be shit. There will be considerable lag. There will be issues due to the monthly bandwidth limits that most ISPs have in place. There will be situations where all of their render machines are filled up, and users will have to wait in line to be able to play a game. This company will be out of business within three years.
An hour of 720p video is typically going to occupy several gigabytes. That’ll chew up bandwidth limits real fast, for heavy gamers.
There will be lag. There is no way to send inputs, render a frame of graphics, encode it into a 720p video stream, and send it back to a user’s screen, without noticeable lag. There is no magical technology they could possibly invent to compensate for this.
Also, while I’m not an expert on the subject of business, I do not think that the business plan is feasible. Unless the service is exorbitantly expensive, then I just don’t see how they will be able to pay for thousands of gameplay rendering boxes (all of which will require maintenance and periodic upgrades to keep up with the latest games), and all the bandwidth, and the expensive special licenses for the games. This makes no sense to me.
Three years, tops. This is an idea so stupid, that I’m willing to rate it as on par with the Phantom on the “All-Time Most Idiotic Videogame Service Start-up Plans.” We will look back and laugh about this, as yet another failed alternative-delivery-method videogame startup; those most of us with our heads on straight, the ones who aren’t falling for the ridiculous headlines about this being “the end of games platforms,” or “the death of PC and/or consoles,” already are.
This just is not viable.
24/03/2009 at 14:08 schizoslayer says:
I have to take issue with this statement:
“Apply this to UT3 as malcolm suggested. UT3 unlike CSS doesn’t actually use client side lag compensation (this being, the client predicting what it thinks will happen vs. what the server decides happens (i.e. css) ) In ut3 a delay of 40ms with no lag compensation is again undetectable.”
As it is just flat out wrong. Unreal Tournament practically invented modern lag compensation. Everything is simulated clientside until such a time as the server has to correct it. If you’ve ever gotten a big lagspike then you will have noticed that everybody keeps moving, turning and firing rather than a game with no lag-compensation which would result in everybody stopping dead.
24/03/2009 at 14:08 I am beginning to understand this comment system says:
It would be awesome to see this technology used in a LAN setting. If you could buy one gaming box and access it from laptops and other lesser boxes around the house that would be pretty incredible.
24/03/2009 at 14:17 Matt says:
schizoslayer, have you played UT3?
24/03/2009 at 14:24 James Tao says:
“It would be awesome to see this technology used in a LAN setting. If you could buy one gaming box and access it from laptops and other lesser boxes around the house that would be pretty incredible.”
On a theoretical level it’s already possible. I’ve been experimenting with gaming between my desktop rig and my laptop using UltraVNC the last few weeks, and while the results haven’t been pretty they have been somewhat encouraging. If I was an absurdly patient man I could have played the strategic map of Empire at around 6 FPS (My damn 10/100 network switch really messed with that), which was roughly the speed I was getting. Is it truly viable for a 4x strategy game? If you don’t mind waiting. Anything remotely realtime? You’d best get on your bike.
I’ll be interested to see if there’s much of an improvement when I get a new router/switch in a few weeks, though. There’s nothing like jury-rigging an inelegant system for an imagined problem to appeal to my mad tinkerer’s side.
24/03/2009 at 14:25 Markoff Chaney says:
Excellent points. I’ll have to amend my original thesis as well. It is viable for Point and Click, Turn Based Card Games (and other low fidelity graphics wise TBS games) and Porn. Basically, anything you can do with one hand where any spike of latency won’t be a real issue. e.g. this service is good for anything that can be done already in a browser with flash with a netbook with no subscription service outside of an internet connection.
Phantom. LOL. That was my first thought when I heard of the Zeebo.
24/03/2009 at 14:27 The Sombrero Kid says:
i’d just like to stress that i don’t think the jist of this is impossible, streaming low definition games with decent lag is possible i’d say, as long as the end user would be willing to accept a reduction in quality over what he can do locally, this would require 1 or 2 things.
the service must be cheaper than the alternative and i believe this is feasible or i’d say that the user would want to run his own server to retain ownership of his software, this model imo lacks the ability to offer the end user something they actually want and takes from them a lot of things they aren’t prepared to give up, also there is no magic going on a free open source equivalent would likely blow it out the water.
24/03/2009 at 14:28 Moo says:
Killzone 2 has an input lag of ~133-167 ms (from pressing a button to seeing the reaction on the TV). There is a lot of controversy around the controls but a lot of people are happy with the game so I don’t see why OnLine couldn’t work given that they distributes servers around the world.
24/03/2009 at 14:36 The Sombrero Kid says:
@Moo
i find it hard to believe killzone 2 sits around for 5 frames (at 30fps) before it implements your input
24/03/2009 at 14:51 Moo says:
There are a lot of youtube videos showing this in slow motion or frame by frame by filming the control and the TV at the same time. For example: http://www.youtube.com/watch?v=b0iFiLK8VVw
24/03/2009 at 14:51 RC-1290'Dreadnought' says:
Funny panic. As if this would kill retail. Its a cool side service. I think it will live next to current services, just like steam lives shoulder to shoulder with retail. Nice tech, I hope you can rent stuff (a few days for a few euros, or a few hours for 2 euros… they could even do promotional play time stuff) like mirrors edge. So you can play the games you don’t want to spend E50 on, like mirrors edge or assassins creed without pirating. Plus, if you are not sure about the next crysis, you don’t have to gamble buying that new GPU just to be able to play the game at high resolution. I am wondering how they are going to keep up that 60 fps in the years to come. Let’s hope their servers are that flexible.
24/03/2009 at 14:53 Pete says:
It’s not a bad idea, provided a couple of things:
- the technical compromises on lag and image quality aren’t too bad
- the pricing reflects the compromises.
I think there might be a market for these games at £1/hour. Not at £40 upfront.
24/03/2009 at 14:53 dsmart says:
And that would be 6mbps bandwidth. For ONE CLIENT.
Yes. Thats rubbish.
24/03/2009 at 14:53 Matt says:
http://www.youtube.com/watch?v=AhVweOBfvU0
Killzone 2 input lag
http://www.youtube.com/watch?v=b0iFiLK8VVw
Killzone 2 slowmo input lag
24/03/2009 at 14:54 treaclepeople says:
There’s a program called StreamMyGame that allows you to access games remotely like this but running on your own rig. It even has a client for Linux on the PS3. It’s mainly used for accessing your gaming rig within your own LAN, but they do allow broadband access. 2mbit upload gives you around 800×600 streamed resolution. There’s a free version, but you need to pay for higher resolutions.
24/03/2009 at 14:55 Matt says:
Actually more prevelantly, read this article:
http://www.gamasutra.com/view/feature/3725/measuring_responsiveness_in_video_.php?print=1
Seems like input lag of up to 200ms is playable.
24/03/2009 at 14:55 Jeremy says:
I think it is very easy to look at any new thing and be bitter and jaded about it, but the fact that there are people even out there trying to create new ways for people to play video games is exciting to me. Sure this whole thing may flop, people might want to have ownership of their games, etc, etc, but at least people are out there innovating. Bitter? Get off your fat gamer butt and try to invent something yourself. Someone asked the question “Are they going to have 100,000 GPU’s just waiting on the other end”, and the answer would be yes, if they had enough subscribers. This whole thing hinges on the play experience, but I’ll hold my judgment back until I can actually see it in action. If this technology actually works like he says it works, then it could be pretty awesome. I’ll hold back my “impossible” statements for the time being, there are a lot of things I take for granted every day that people thought were impossible 30 years ago.
24/03/2009 at 14:56 Andthensobecause says:
Until the internet grows up and becomes more responsible, I am not going to let it “hold onto” my things “so that I can access them whenever I want.”
24/03/2009 at 15:00 Irish Al says:
@framstick
Logmein (and GotoMyPC, and Webex and VNC et al) are all very fine and well for remote desktop access, because they work in a lower colour depth and only update screen regions that have changed. In an action game the whole frame will change most of the time. If the entire internet was rock-solid stable gigabit speed from end to end then it *might* have a chance. That ain’t happening anytime in the next 30 years.
24/03/2009 at 15:08 skalpadda says:
Jeremy: It’s certainly an interesting and commendable idea, but I think most people (especially those here) are quite fond of their PCs and like running games on them. The need to go out and invent something yourself is just not there since the way it currently works is (mostly) just fine.
In the hypothetical situation that this turns out to really work and be a sustainable business I’d certainly find it interesting to be able to try out very flashy new games without having to upgrade my hardware. On the other hand I very much like the way I’m enjoying games on my hardware now and don’t really see enough advantages to running games locally on my own machine.
And this is all assuming the tech actually works of course :)
24/03/2009 at 15:13 Xercies says:
12.5 fps to me isn’t that bad, that was the average i got for Crysis on medium and it didn’t affect the game that much. Actually yo can get down to 10 without it affecting much which is what usually my games go down because i have a rubbish system.
24/03/2009 at 15:14 suibhne says:
@Matt and others talking about ping: you seriously don’t notice 40ms of ping? I’ve played FPS games competitively for about years, and I clearly notice differences of even 12-16ms – and I’m not even that good. 40ms is huge in any game with hitscan weapons.
Don’t forget, too, that you’re just talking here about latency that would normally be client-side – so the 40ms latency from OnLive is in addition to the latency of OnLive connecting to a MP server, adding another 12-16ms even if they’re located in the same city (and probably even if they’re in the same datacenter). Even if you’re one of those gamers who would never notice 40ms, 60-80ms starts to look a little different.
24/03/2009 at 15:17 dsmart says:
@ Matt
Those videos are woefully inconclusive. The implications are that they gimped KZ2. And the reason for that would be what exactly? Its not like they’re so starved for fps that they’re going to save cycles by not doing frequent input updates.
24/03/2009 at 15:20 suibhne says:
Also in response to those folks batting around input lag figures: again, as I mentioned above, network latency from OnLive would be [i]on top of[/i] input lag already in the game – and would be effectively doubled, too, since your input would have to travel via network, where the game would respond in whatever laggy manner is coded into it, and then the game’s response would have to travel back to you via network. And that assumes there’s no additional latency from client-side video processing for OnLive.
24/03/2009 at 15:21 nakke says:
Seems like total vaporware the way they’re presenting it. Where the heck are they going to get the processing power for tens of thousands of simultaneous players?
24/03/2009 at 15:23 Black Mamba says:
While I suspect the the high end PC gamers and tech fetish crowd will hold their nose at such a concept this could be pretty big deal for those who dont care about tech and want it to just work.
Plus the explosion in netbooks and the fact that it works on the Apple platform as well means an big chunk of people out there could happily play games through the service assuming it works.
Their little device which plugs into the TV enabling the service is pretty neat as well and could certainly hurt the games consoles too.
24/03/2009 at 15:24 Iain says:
@dsmart:
The last time I saw/heard about something remotely like this, it was called the Phantom console.
Yup, I have to say this does bring on strong memories of the Infinium Phantom – the biggest videogames industry vapourware joke since Duke Nukem Forever. Years and millions of dollars of development, and all they have to show for it is a lapboard.
If I recall, their model was to stream PC games entirely down to an under-the-TV console, so that was never really going to work. The console would have been obsolete in seconds and most people’s internet connections were so slow that downloading a game before you were able to play it was a non-starter. (and this was based on a rent-per-evening model!)
Of course, cloud computing is an entirely different kettle of fish, but I’m struggling to see how it could be applied to games, where fast input-output responses are critical. If you look at the state of the broadband infrastructure in the UK, it’s pretty much uniformly terrible.
I think think this would be a great idea in about 5 or 10 years, when we might actually stand a chance of getting a better infrastructure in place (British Telecom have had fibre optics technology for well over 20 years – they’re just too fucking tight-fisted to do a nationwide rollout, which is why our average broadband speed is still in the Dark Ages and roundly shamed by most of Eastern Europe)
As stated upthread – I think the ’1ms latency’ statement is deliberately disingenuous. He’s talking about the video rendering lag and hoping that people interpret it as ‘there will be no lag at all’, which if you have any experience of online gaming, you know is utter crap.
Anything under 200ms will be good enough to play most things, but beyond that and you’re really going to be testing the customer’s patience and forgiveness.
24/03/2009 at 15:25 Tei says:
The FPS games feel smooth on the internet, because the client fake the server status with calculations. Everything on the screen (your position, the other players positions, the rockets) are predicted. Withouth prediction, you will be playing with 400 ms lag. Is really hard to play with 400 ms from your action to the screen, you move like you are drunk. 400 ms is not your typical 400 ms lag on a game withouth prediction.
Having a game streamed to you has video == zero game prediction. Your inputs will just be send to the server, with not local calculations at all. So you are stuck with a “outdated” image on the screen, till the server receive your input, process a new image, and you receive the new image. Who knows? maybe 300 ms later.
I would love to see a service like that for no-gamers. I think the “TV digital” sould have ben like that, and not the totally stupid stuff we call nowdays “TV digital”. But for FPS games IMHO, will suck.
24/03/2009 at 15:25 Matt says:
dmsart: read the article I linked to
suibhne: The assertion is that the input lag I’ve linked to is specifcally for killzone2. In general PC games don’t have this input lag, so the input lag wouldn’t exist in the games we’d be playing over the system. Therefore we can compare the network delay of onlive with the input delay of killzone2 as it boils down to the same result.
24/03/2009 at 15:28 teo says:
CSS is shit, maybe that’s why you don’t notice it
Try playing CS 1.6. At 40 ms starts getting stupid, and if you’ve ever played with 10 ping you’ll never want to go back. There is a difference, too bad if you can’t tell.
Also, 200ms input lag? Are you fcking kidding me? That’s why I don’t play console games. When I pick off people in mid on dust 2 my reaction time is less than 150 ms, and you don’t think I would notice 200 ms input lag? rofl
24/03/2009 at 15:31 Tei says:
“400 ms is not your typical 400 ms lag on a game withouth prediction. ”
long version:
I mean, 400 ms on a game with prediction is playable. Maybe you will fail your shot, and everyone will pwn you. But you will be still able to navigate the map, pick weapons, maybe even pick weapons that are not here.
400 ms in a game withouth prediction mean you move like a drunk bastard, and is really hard to move from point A to point B on the map.
24/03/2009 at 15:34 Jeremy says:
Suibhne, I agree with you on the lag issue, but for me it kinda works in levels. For instance, 0 – 20 seems fine, 21 – 50 I get pissed a couple of times and cry “LAG” but at 51 – 100 I just hate the game, it becomes very unplayable to me. I was spoiled at college on a T3 line and had an average ping of 15ms on the counter strike servers I played on.
Skalpadda, I am 100% a PC gamer but there are those console titles that I really would love to play but don’t want to shell out the cash for the console itself. If there was a way for me to play those console titles without purchasing the console, then this could be worth it for me.
Also, we may like things the way they are, but what if some innovation makes it better? Everyone was happy with riding horses until people started making cars. Everyone was happy with driving cars until some idiot thought he could make a machine that flies. This isn’t quite as drastic as that of course, but innovation leads us to amazing things.
24/03/2009 at 15:48 Iain says:
To all those people complaining about latencies of 50ms or more being unplayable… man, I’d love to see you on a 128 player server of Joint Operations on a 1Mbps line, I really would.
24/03/2009 at 15:58 The Sombrero Kid says:
i’d just like to say that a 167ms input response means the updates are being REALLY stalled, since i doubt they update less frequently than they render (this is NEVER done) this would require frame rates sub 5fps AND 90′s style code, it’s much more likely you’re getting triple buffered imaging resulting in a 2 frame response at 20-30fps and a bunch of hysterical people who don’t know what the hell they’re talking about.
24/03/2009 at 16:00 Skurmedel says:
If something is limiting I think it would be the costs involved, and the processing power needed on their part. I don’t think the bandwidth is a real issue. I have digital tv streamed to my set-top box. My ISP reserves about 5 mbit for this, and it works like a charm. In total I have about 14 mbit in, 8 – 9 while using my TV and I can download things at very reasonable speeds and play games. Granted, this is high-quality SDTV. But I do think streaming the images is a non problem when people start getting 20 mbit inbound bandwidth and higher provided their servers won’t be situated in Somalia or something.
24/03/2009 at 16:07 jalf says:
Oh god, what are those publishers smoking?
It’d obviously be the wet dream of any anti-piracy fetishist, but apart from that, it just doesn’t make sense.
1.5mbit minimum, eh?
Now let’s say I don’t actually want to go back to 1998 era graphics. That means I want to be able to continue playing at 1600×1200. It also means I don’t want jpeg compression artifacts. So png is probably the most sensible way to transfer the image for each frame. I just so happen to have a rather simple image here (black&white only, and very little visual noise, should be ideal for compression purposes). Its size? 1.17MB. That’s megabyte, not bit. In other words, to transfer *one* such image in one second would require, ohh…. 9.36 megabit/second.
Now, what is the absolute minimum framerate we could tolerate in a slow, non-twitchy game? 20 FPS?
Great, a mere 188mbit/second should do the trick then!
And that’s without sound.
Then there’s another slight problem. Rendering time doesn’t just vanish just because it happens on a remote server. Even a relatively beefy computer is limited to, say, 60fps because rendering takes around 1/60th second, or 17 milliseconds. There’s no reason to believe their server farms could do it faster. (What, are they likely to buy top of the line quad SLI hardware for *every* customer?).
Then we can add the network latency on top of this.
That is, if you try to run Crysis, you go from a latency of, say, 17ms (pure rendering time) to, say, 77ms (assuming 60 ms network latency, which is about what I get to google.dk, I’m on a 20mbit line, and I live in the middle of the biggest city in Denmark, so this is an optimistic estimate. The world is a big place, they’re not going to give *everyone* 60ms coverage.
You think Counterstrike is playable at ~80ms lag? Good for you. That’s because your client software tries to predict and compensate for the lag. The server can hardly do this, since it doesn’t know how much you’re going to lag. You’re going to see what happened 3-4 frames ago. That is more than enough to feel *very* odd when playing.
@Matt and others: Input lag exists on every system. It is supposed to be particularly bad on some consoles (I’ve heard 360 was especially bad), but Windows loses a frame or two as well, waiting for the input to get from your mouse to the CPU as well. And of course, this doesn’t go away *either*. We can add it to the growing list of latency. And while we’re at it, the refresh interval for your monitor. 60hz? That means some quite long wait periods if the image data doesn’t arrive *exactly* synced with the monitor. Which of course, it doesn’t.
While input lag of 200ms may be playable, once you add everything else on top of it, you’re looking at a total of maybe 400ms. Which, you know, doesn’t seem like an improvement to me.
@Jeremy: I fail to see the innovation though. If they could let me run my games without having to pay for the hardware, I’d love it. But that’s not what they’re going to be able to do, even if they wanted to. If the alternative was “waiting around until a fairy came and teleported you to the place you wanted to go”, then I daresay people would *still* be happy riding horses. The car only took over because it was a realistic alternative. This is not a realistic alternative.
I don’t want to play laggy versions of 1998-era games in 1993-era screen resolution. I don’t know about you guys. Maybe I’m just picky.
But of course, why wouldn’t the publishers “sign up” for it? It doesn’t cost them anything. If it takes off, they’d be happy to use it. And if it doesn’t, eh, what a shame. Good thing we hadn’t yet spent $100 million developing games for the platform, eh? “Signing up” doesn’t mean “we’re going to port our AAA titles to the platform. If they were going to do that, they’d be insane, and probably out of business in a year or two.
24/03/2009 at 16:08 schizoslayer says:
@ Matt not only have I played UT3 but I’ve been programming in Unreal based engines for roughly 5 years total. I am incredibly familiar with how the replication works and the entire engine is based on client-side simulation with the server providing updates that ensure the players are where they should be.
A game with no lag-compensation sends your input to the server. The server moves you and tells the Client to move your avatar. UT lets you move around client-side and then adjusts your position to match the server if they are too dissimilar which removes the input lag.
24/03/2009 at 16:16 schizoslayer says:
Where has Edit gone? Bah. It removes the input lag that would result from sending your input to a server. Input lag inherent in the workings of your PC still exists.
24/03/2009 at 16:18 Mad Doc MacRae says:
Not a lot to add other than by the time they get this working, indie developers will be doing something even cooler.
24/03/2009 at 16:23 The Sombrero Kid says:
incidently decent pc kit would have input lag of 3ms or there abouts then theres frame processing and swapping and if it’s triple buffered there’ll be another set of rendering and processing before you see the results at 60fps input lag on a pc of the last 5 years would be 20-30ms tops.
24/03/2009 at 16:25 teo says:
jalf they were talking video compression not image compression
24/03/2009 at 16:29 The Sombrero Kid says:
@teo
lol video compression are you mental! holding back frames to compress the video up would introduce a very interesting amount of lag.
24/03/2009 at 16:38 Alvis Brigis says:
If verified, this compression technology will revolutionize not just video game distribution and play, but also virtual worlds, mirror worlds (Google Earth), HD WebTV and so forth. Wow.
http://memebox.com/futureblogger/show/1730
24/03/2009 at 16:38 Cunzy1 1 says:
When I use chart wizard in excel my reaction time is 4(pc)ps. Using this system I might as well draw them by hand. ON PAPER.
24/03/2009 at 17:01 Matt says:
Ok, using Sombrero’s 30ms limit for input lag on a computer we’ve got lag in total of:
80ms network delay
30ms input delay
1ms video compression delay
8ms sync delay on average (Frame to 60hz)
= 119ms
If the ps3 is playable with an input delay of 160ms, then 119ms will be more than acceptable.
24/03/2009 at 17:02 Matt says:
Edit:
+ 16ms delay for the actual rendering of the frame
= 135ms
24/03/2009 at 17:16 suibhne says:
Not to be a platform bigot about it, but one of the reasons that such terribad lag might be considered “playable” on PS3 is that dual-analog control can’t even dream of someday imagining the possibility of growing up to even live in the same region as the aiming precision of a mouse.
24/03/2009 at 17:30 jalf says:
I’m not sure what Sombrero bases his 3ms input lag on. It is far above that. 50ms is probably about the best you can get on a PC as well. (Just waiting for Windows to interrupt your game so that it can process the input may take up to 10ms. Windows isn’t constantly running in the background).
and one of the reasons why PS3 has such high input latency is that the application isn’t constantly checking for input either. Even once the signal has been sent to the application, the app has to wait until the next frame, where it again checks for input. And of course, one reason why 160ms might be bearable on the ps3 is that most games run at 30fps, not 60. So it’s only half as many frames we see pass before our input registers. On a PC you typically expect 60fps or higher, so twice as many frames are rendered in the same amount of latency, which contributes to it feeling sluggish.
You’re also forgetting that rendering a frame takes more than 16ms once you include checking for input and processing game logic – and sending data to the gpu. http://realtimecollisiondetection.net/blog/?p=30 may be worth a read.
The PS3 isn’t badly designed. They didn’t just add 150ms latency for no reason. That’s just what it adds up to all in all, under any circumstances. Even on the PC, the numbers are similar. And now we’re adding network latency on top of it.
24/03/2009 at 17:43 Lobotomist says:
Great!
Unfortunately its just a dream :D
24/03/2009 at 17:45 Jeremy says:
Jalf, true, nobody would want to wait for a fairy when they could ride a horse, but I hardly see what relevance that has to this conversation. Innovation of thought, innovation of technology, all of these things can lead to bigger and better ideas, even if the original plan fails. So you’re telling me that being able to play any game from any platform without owning said game or platform isn’t innovative thinking? How many other failed attempts at transportation were there before the car took off? How many times did the plane fail before it finally worked?
24/03/2009 at 18:03 Matt says:
Jaff rendering a frame can’t take more than 16ms if it’s rendering at 60 frames a second.
Assumign you’re right about the 50ms input lag, then that’s still a totally delay 155ms and the ps3 is at 160ms.
If the ps3 can be playable at 160ms then there is no reason why this system at 155ms wouldn’t be equally playable.
I’m not sure the number of frames you see a second contributes to whether it feels less or more sluggish. You still see the response to your input in the same amount of time, regardless of whether you see 3 frames or 6 frames before that time.
24/03/2009 at 18:17 rnx says:
with servers hosted at your ISP or nearby and somewhat limited resolution this could certainly work and work profitable.
having expensive gaming hardware in every household that is not used most of the time is a huge waste.
on top of that they can use hardware without the heat/noise etc. requirements of consumer devices.
24/03/2009 at 18:18 Matt says:
http://www.n4g.com/events_gdc2009/News-299332.aspx
2nd video describes pricing,
Wireless controler “described in latency terms of microseconds *not* milliseconds”
Scheduled to launch winter 2009
“Server centers are designed with extremely low latency considerations, we couldn’t just drop the onlive system into a regular data centre and make it work”
24/03/2009 at 18:29 zak canard says:
I’m going to sit on the fence until some valid proof of it in action materialises from a trusted and reputable source. If they really want to silence the sceptics they also need to make the compression technology being used available to all and sundry, preferably for nuppence.
24/03/2009 at 18:42 dsmart says:
This is just a load of Gaas*. It will never fly. Ever.
* sorry, I couldn’t help myself. Fine, go ahead, shoot me :)
24/03/2009 at 19:25 schizoslayer says:
To use some non-theorycraft numbers I just grabbed from UT3…
Input Latency was roughly 23ms (occassionally it spiked up as high as 50ms) Ping to a dedicated server in the UK was 57ms.
You can find these stats in UT3 easily with the commands:
stat engine
stat net
So from Hand to Game it’s 23ms. From Game to server it’s 57ms.
Your mileage may vary. The harder your PC is being stressed the worse the Input Latency gets.
However imagine this:
Playing a game of UT3 via a Cloud version of UT3.
You not only have the trip time between your PC and the Cloud but then you have the trip time between the Cloud and the Server.
As a quality experience there is no way for this service to compete. The only way it can compete is in non-time critical games with a market that isn’t interested in graphical fidelity or responsiveness.
24/03/2009 at 19:29 mashakos says:
“Ok @ all the people worried about lag.
A.) When is lag discernable?”
ouch. You fail to realise that CSS is played locally, your input / sound / on screen gfx are all processed locally. When everything is done remotely, you WILL notice a 1 second lag between your hand moving your mouse and the view rotating on-screen.
Keep on dreaming guys.
24/03/2009 at 19:48 TheSombreroKid says:
@jalf
sorry but your wrong about this, no game written in 10 years has used windows api’s to read the input from the driver, they use direct input or similar hihg performace api, given all microsoft mouse have a latency of 3ms and most gaming mice i.e. the ones gamers have have a sub 1ms response, the time taken for the input to reach the driver is at most 3 ms, once there it will wait till the update loop is called but this is not linked to frame rate and again hasn’t been for at least 10 years this information will be read before the next frame is written which will be at most 16ms, then the frame which is based on this information will be written 16ms after that in a double buffered system thus the response time will have an absolute ceiling of 42ms given a fixed framerate of 60fps, this i’m afraid is irrefutable.
also in 60fps system the time between frames is 16ms not the time to build the frame, that vastly depends on the scene.
triple buffering is very commonly applied on th e ps3 because it’s extreamly hard to get rid of the tearing, i’m certain i read a gamasutra tricle with the killzone guys that says they use it, this combined with the bluetooth latency issues would result in an absolute ceiling of more than double the PC at around 100ms (32ms per frame *3 + 10ms latency on bluetooth)
i’m not sure what the latency on the bluetooth controllers are but i’ve heard it’s not great, perhaps this could be the source of the missing 60ms, maybe it’s that bad.
24/03/2009 at 19:50 TheSombreroKid says:
slight disclaimer:
with bad prgramming you could easliy increase your latency, reading input at the end of an update instead of hte begining and things like that :D
24/03/2009 at 20:05 Tei says:
“80ms network delay
30ms input delay
1ms video compression delay
8ms sync delay on average (Frame to 60hz)
= 119ms”
+ downloading 800 KB.. say …400 ms (ymmv)
+ uncompress the data, update the screen, , screen retrace. 20 ms (ymmv)
= 719 ms
Good luck playing a FPS with 719 ms of latency and zero prediction.
24/03/2009 at 20:13 Jeremy says:
Good thing you didn’t design it ;)
24/03/2009 at 20:21 Matt says:
@Tei:
They claim that they can display 60fps, this implies that the download delay for a single frame can only be as large as 16ms, so the 400ms figure is irrelevent.
As for uncompressing the data, their compression algorithim is 1ms, their uncompression algorithim will be < 1ms.
I don’t know where you’re getting the numbers for how long it takes to update the screen, however, again, given an fps of 60, it’s a absolute maximum of 16ms.
So taking all that into account it’s a maximum of:
127ms + 1 + 16 + 16 = 161 ms, which is about the same as the ps3, so again, should be fine.
24/03/2009 at 20:26 TheSombreroKid says:
frame rate restriced by bandwidth, lag is restricted by latency you can have 4 million frames per second with 4 million millisecond latency if you want.
24/03/2009 at 20:35 Matt says:
True, but considering we have no information about the size of a compressed frame, it’s difficult to make any assumptions about it.
24/03/2009 at 20:37 Jeremy says:
People who have used the hardware have already claimed that it works well, if not as smooth as the actual hardware, it is very close and most definitely playable according to all cases I have heard thus far. The true test will be in the homes of actual users I suppose, but there is an invitational beta showing up soon, maybe one of us RPS folks would be lucky enough to get an invite.
Also, keep in mind that Perlman is no fool, he’s been around for a long time and has developed several very cool bits of tech in his day.
24/03/2009 at 20:43 Matt says:
Actually thinking about it sombrero you can’t (well you sort of can)
You can have an infinate amount of end to end latency, and have 4million frames a second with an infinate delay from when the first one arrives.
However, in order to display 4million frames a second, each frame must arrive consequently 4millionth of a second after the first one, regardless of lag. So the amount of time it takes to download the packet of whatever size that is (regardless of latency which we’ve already confirmed as 80ms) is 4millionth of a second
Scaling this to 60fps = 16ms
24/03/2009 at 20:48 Jocho says:
Anyone remember the days when people said stuff like “The Internet is a fad”?
24/03/2009 at 20:55 Jeremy says:
I still don’t believe in the internet. Al Gore said he created it, and quite frankly, that debunks it in my book.
24/03/2009 at 21:07 TheSombreroKid says:
@Matt
yep as i said before frame rate is dependant on bandwidth NOT latency
24/03/2009 at 21:15 Matt says:
Sorry, I’m failing at reading today.
24/03/2009 at 21:19 Tei says:
“Anyone remember the days when people said stuff like “The Internet is a fad”?”
You can use these words to accept *anything*. Have a too close mind, and you die out of oxigen, have a too open mind, and you brain fall to the ground, and you steop over it.
24/03/2009 at 21:22 Mike says:
This will work, but I don’t think we’re at the point of actually needing it yet.
24/03/2009 at 21:28 espy says:
If it works, I’m in. Brilliant idea, and the whole concept suits me wonderfully.
24/03/2009 at 21:37 TheSombreroKid says:
my argument all along has been just because we can do something doesn’t mean we should basically
24/03/2009 at 21:40 dsmart says:
No. It won’t work. Now can I have the last word? Please? :)
Oh, and btw, EVERY SINGLE ONE of these calculations you all have going here, get a D- because you guys are all well and truly off the mark because without two bits of information, you simply CANNOT do the math.
1. The size of each frame being sent
2. What happens to the frame upon receipt? Does it get decompressed, rendered as-is, shoved into a local FILO buffer?
etc etc
24/03/2009 at 22:00 Moorkh says:
Hmmm… not only am I rather convinced this will fail at this time, I very dearly hope that this monster will never work nearly as well as games running locally. I’m afraid, though, that it will, eventually.
‘Cause whenever this comes to pass, even console gaming, which I loathe and have sworn to never practise, will suddenly seem like a perfectly sensible and mature way to play games. Of course, neither PC nor console gaming would long survive this development aside from some niche and non-commercial products.
The mainstream would instantly be won over by such a system and freely give up their control and ownership of their game, as they just want to have some shallow fun with it. Publishers would love to reduce their service to just that, avoiding mean things such as pirating, modding or any responsibility towards the player. Both console and dedicated gaming PC hardware would quickly be too expensive to even develop, much less sell due to massively reduced demand. Naturally, reducing their services to blockbuster games is a logical next step for the new service providers, since a streamlined game library would increase flexibility and reduce maintenance for their server farms.
What remained locally would be independent PC gaming with low technical requirements. Not gonna see DF on no OnLive, to be certain. A candle of hope remains.
24/03/2009 at 22:00 Oktember says:
One thing they’re underplaying is the resolution these games will be running in. 720p is the current maximum over 5mbps in a lab. That’s 1280×720. Welcome to 2002.
Anyone connecting at less than 5mbps will get the 480p standard definition experience. That’s 720×480. Welcome to 1994.
I’m not quite sure why this is being orgasmically hailed as the future of PC gaming. From where I’m sitting, it’s revisiting the dim, distant past.
24/03/2009 at 22:01 Jeremy says:
Except it works. Do that math.
24/03/2009 at 23:05 dsmart says:
@ Jeremy
lol!! uhm, no it doesn’t.
24/03/2009 at 23:09 RC-1290'Dreadnought' says:
Oktember, you don’t have a 20Mbit connection I guess?
24/03/2009 at 23:21 Matt says:
lol dsmart.
So you claim it won’t work, despite the fact that, the company has clearly invested a lot of time and effort, there are reports of people using the system and it working, and you also have no knowledge of the size of the compressed frame.
You’ve made a sweeping judgement. All the input lag consideration has been without knowledge of the size of the compressed frames; I’m not claiming it will definitely work because I don’t know what those numbers are. However you’re claiming it won’t work with exactly the same knowledge.
TBH more intelligent people than either of us have spent 7 years developing this system, given that they’ve demonstrated it to press, suggests very strongly they’ve got it figured out and working.
I’m actually quite shocked that as you are described as a “prominent industry player and regarded as one of the quintessential indie developers” you’re making these remarks.
By the way who wrote your biography?
24/03/2009 at 23:57 ChampionHyena says:
Anyone getting PowerPlay flashbacks?
25/03/2009 at 00:18 schizoslayer says:
In a world with an internet with infinite bandwidth making the size of a single frame irrelevant you can’t beat the Speed of Light.
The time it takes for your input to be displayed back to you is dependent on how far from the datacenter you are. It has nothing to do with how fat your pipe is. If you live 1 mile down the road from the DataCenter then you will be fine. If you live 1000 miles from the Datacenter then you’re screwed.
Given that the internet isn’t even close to approaching the speed of light thanks to all the garbage cluttering it up and the amount of hardware your packets have to traverse before they get to you (not forgetting that your ISP will be inspecting all of those packets to make sure you aren’t being naughty) you’re lucky if you even get to 50% the speed of light.
Playing Company of Heroes 15 miles from the Datacenter over VPN (which is how it has been demoed in the past) is very different to playing Guitar Hero 1500 miles from the DataCenter over the public Internet.
I’m not saying it can’t work. I’m just saying that the quality of the service isn’t going to come close to unseating consoles or home PCs.
25/03/2009 at 00:46 mashakos says:
the guys rooting for this are hilarious. Why don’t we all hope for lag-free HD youtube streaming before we jump to streamed games?
…and yeah, I’m not exactly amazed by the proposed 720p resolution when I game at 2560×1600
25/03/2009 at 00:47 mashakos says:
onlive should have promised something more realistic. Like economy class tickets to the moon.
(wish I could have said that in my prev post >.< )
25/03/2009 at 01:08 Robin says:
“So you claim it won’t work, despite the fact that, the company has clearly invested a lot of time and effort, there are reports of people using the system and it working, and you also have no knowledge of the size of the compressed frame.”
They invested a lot of time and effort into the Titanic.
25/03/2009 at 02:02 Andy`` says:
Keep seeing this kind of idea pop up on Slashdot occasonally, the most recent being something to do with ATI (might have been a patent?). Obviously, it’s followed by a string of comments explaining how its a stupid idea. I don’t see it as impossible, but there’s so many things (most covered by people above) that make it a fair bit challenging to do well.
Personally the first problem I see with it is having an internet connection that’s a) capable of making the service with paying for, b) doesn’t cost the earth and c) is stable enough so that I’m stuck without gaming. Aint going to happen here. Hong Kong will be ok though: http://www.netvigator.com/netvigator2003/serviceplans/service_plan.html#bb
Personal economics of its customers will seal its fate in the end, one way or the other (an unfortunately bad financial climate to be releasing things like this too!)
25/03/2009 at 02:05 Andy`` says:
Worth paying for, not with paying for.
Though, is it remotely possible that net cafes can make use of the service (and it be viable)?
25/03/2009 at 02:39 Anthony Damiani says:
My understanding is that this approach is too bandwidth-dependent to be actually viable.
25/03/2009 at 02:47 Sid Clayman says:
I worked for a small company which provided this service to other companies. Except it wasn’t games, it was desktops and any/all applications required.
Great in theory. I loved the idea of nobody needing a decent computer for anything, you just stream it from the servers.
With our small company there were tons of problems however, and speed was really inconsistent. And our servers were sometimes dodgy. Having MS Word and Outlook lagging annoys people. :(
It’s also expensive. Everybody will surely have to pay a subscription fee to the guys who own the servers providing the service
(even if this done come to pass, the option will always remain to download the games to your own hardware)
25/03/2009 at 07:14 SteveHatesYou says:
Here’s a good article on response lag, for anybody who’s curious: http://www.gamasutra.com/view/feature/1942/programming_responsiveness.php
Typical lag with 60 fps is 50ms. 30 fps is 100 ms. Keep in mind that there’s three frames between you pressing the button and then seeing the result.
So, 80 ms, if it really is the *total* response time, would absolutely be playable. And if they really have pulled this off, it could be a massive shakeup for the industry.
25/03/2009 at 10:28 tentacleraep says:
The only way this could work today was if the actual rendering was client side and that would mean everyone would need a beefy gfx-card anyway. All of you talking about video compression instead of image compression are missing the fact that most of the efficient (size wise) video compression methods use keyframes and morph the images in between, this also takes quite a bit of power and I doubt that they will have the power to spare since you will need to almost double the required power per stream to be able to compress it in a playable speed.
The problem with this in games is that keyframes are best choosen based on what the coming frames look like so you don’t take an almost blank frame as a keyframe when the next will be a frame with a lot of detail or a lot of changed details. Whenever lots of details change you will need a new keyframe that will be basically be a fully rendered frame.
Most of you have been talking about the video aspect of the problem and the input lag, the audio will be a big problem since sound, in good quality, will take a lot of bandwidth, this could be solved with having the audio client-side, which means that you will still need to install the game.
Another problem that might arise is lag-spikes, the reason that streaming audio and video works well is because you can buffer it, most mediaplayers that support streaming have a buffer of several seconds so that you won’t have sudden pauses because of lag-spikes. Lag-spikes in this system will either make you stand still or it will make your screen freeze.
25/03/2009 at 10:52 Echo says:
Where will OnLive be available?
We’ll be launching across the continental US.
:(
25/03/2009 at 11:19 Matt says:
tentacleraep.
In regards to the video compression, they claim their compressor compresses frames in 1ms. If this is true then it’s a non-issue.
As for audio, audio takes significantly less bandwidth than video. E.g: 128kbs compression of the game audio will mean that it’s around 12kB per second which is really nothing and provides perfectly acceptable sound.
25/03/2009 at 11:34 Jesucristo says:
If some day this works, What games will be published?. Could be see games like Armed Assault or DCS Black Shark?.
I think this is only a new piece of shit to obey the players to play the games the editors want.
25/03/2009 at 11:45 dsmart says:
@ Matt
OK, this is just not funny anymore.
Do you know how many companies invest lots of time, effort and money on stuff that just.does.not.work? Its called R&D most of the time.
I don’t care how many people are in focus testing, unless it is out in the wild and under real-world conditions, there is no evidence that it works.
Phantom, DISCover and everything in between have tried this and failed. Why? For the same aforementioned reasons. Unless you’re playing peggle and are happy with 1999 resolutions, sure it will work.
You think it will work? Fine. Go try this then. Report back.
http://www.streammygame.com
25/03/2009 at 11:51 madhaha says:
Isn’t this essentially how SecondLife works? And didn’t we learn that a) it was laggy as hell b) didn’t scale c) looks like arse and d) still requires a beefy machine.
25/03/2009 at 12:24 Matt says:
- Yes because obviously all those companies have used identical technology.
- If by “1999″ resolutions you actually mean higher than the average resolution @ 720p (http://www.screen-resolution.com/common-screen-resolutions.php) then yeah sure.
You’re right, there’s no evidence it definitely works, but that’s definitely more evidence suggesting it works than evidence suggesting it doesn’t. Again, I’m not claiming it definitely works whereas you seem panic driven to demand that it “will never fly, ever” with all that…. erm…. “proof” you have.
P.S. I’m still interested in knowing who wrote your biography?
25/03/2009 at 14:12 nikos says:
RE the notion that a FILO (sic) buffer might be used to store frames before rendering: either this is a typo, or an ingenious way to play computer games through a memento-style filter….
10/03/2010 at 20:27 Leesping says:
@ Nikos: It’s probably not a good thing that I found your post funny. :)
25/03/2009 at 14:30 Darren says:
Guys, this will work. Why? Because they have algorithims. And not just any ol’ algorithim – COMPLEX algorithims. The answer to all of life’s problems.
Just like defrag is the answer to all of computer errors. Now, if they just use a complex defrag algorithim – holy crap!
25/03/2009 at 14:43 Jeremy says:
I guess all of those articles I read about games journalists playing games using OnLive were all bogus. Fabrications. It is probably some huge conspiracy with the end goal being to destroy gaming for all time.
25/03/2009 at 14:49 cassus says:
This is a brilliant scam, and the perfect example of how easy it is to fool dumb publishers into forking over cash. Times must not be all that tough on publishers as they make us believe. “buhuu, piracy bla bla bla. Oh you can make zero latency connections you say? Here’s 50mill, hope you make some advances on your faster than light connectivity! It will solve all our problems. Now we just need a renderpark that dwarfs any supercomputer cluster ever made! WE’LL ALL BE ROLLING IN DOUGH!!!!”
Fricken idiots. For each guy playing, you’ll need a freakishly hefty gaming rig, an awesome upstream to that client, his downstream has to be more than decent, and packetloss free (packetloss = screwed) and you’d probably need fiber all the way to your computer for this to even seem close to possible. That leaves out about 99% of all possible subscribers. Unless this is actually technology that’s ready around 2025 or so. In which case, stop hyping it.
Fricken faster than light latency.. Gotta love the idea.
Bet ubisoft and EA is all over this with huge amounts of cash. Salivating and panting like frisky little puppies.
I don’t even see why anyone would want this. Seriously, what’s so great about this? “oooh you can play games the way you play them off your harddrive, but all the graphics and sound and whatnot will go through the interwebs!! ZOMG! FANTASTIC!”
Going to be awesome for people in the UK with download limits on their intertubes. Get to play 4 minutes a month. It’s so evident that this would only be useful for people with 85 year old computers. And if this actually did succeed, intel and amd and nvidia and so forth would just die out. Cause they would just be making hardware for workstations at that point. You can bet your ass intel won’t stand for that kinda crap. I’m also pretty sure Intel isn’t all that concerned about this type of alchemy.
Dumbdumbdumbdumb-duuuumb.
25/03/2009 at 14:56 cassus says:
I would also like to add that the 720p compressed image would render any kind of tiny detail worthless on a LCD screen that’s not 720p native. And i think the last 720p native LCD screen went off the market when the amiga 1200 stopped being produced. Or thereabouts.
Yesterdays technology – TODAY!
25/03/2009 at 15:09 Aldo says:
25/03/2009 at 15:10 Aldo says:
Ooops, forgot to close the quote…..
25/03/2009 at 15:25 cassus says:
Aldo said:
“After all, since when did customers drive services?”
Customers can drive services down the drain. If someone develops sub par rubbish and the customers go “yeah.. good work there” and moves on, it’s sleepytimes for that product. Just look at all the MMO’s that don’t listen to the beta testers and the general public. People have more say in what’s successful than anyone with cash really cares to admit. Sure, this thing might sell to suckers who don’t care about rubbish graphics (1080p might seem like xanadu to console gamers, but consider the fact that it’s compressed. and at 1.5mbit it’s compressed REALLY hard. Now your wife or sister or whatever wants to download some episodes of sex and the city from a torrent site..
This tech is seriously just dumb. So many silly things in the way. If you live outside of a major city, this is going to be completely useless. Even more so if you live in a country where bandwidth is limited and you have a set amount of gigs pr month and so forth.
It’s gonna sort of work in the US, and it’ll work in certain parts of asia. Forget northern europe, the countries that were the former soviet union and most of eastern europe.
Games will still have to be downloadable or store bought on a physical medium. And that version would be superior to any streamed crap because it would be realtime (no latency) and it would have sharp nice picture quality and brilliant uncompressed sound. So i say again.. Who would want to pay a monthly fee for this? There are streaming game sites right now that stream games to your computer and everything plays local, but no one uses them. Cause it’s inconvenient and crappy.
This might be a little bit better, if it works the way they say it will, but it will also be more expensive, and also available to a way smaller userbase.
Anyone thinking this will stop piracy as a twat, and that’s the only reason for anyone to actually want to publish through this. People who can’t afford to upgrade their pc’s once a year for gaming performance already bought a console. And all games these days are released on multi platforms anyways.
dumdedumdum-silly.
25/03/2009 at 16:37 espy says:
The amount of callousness in this comment thread is amazing 0_o
The amount of misinformation and idiocy equally so. Nobody ever said anything of ftl transmissions or similar nonsense.
Here’s today’s presentation at the GDC, one hour long, they play Crysis on a low end Dell, a Macbook and the Microconsole via the actual Datacenter that will host the finished service as well.
Plus there’s a Q&A at the end where they talk a lot about lag, bandwidth etc.
http://www.engadget.com/2009/03/25/video-onlive-streaming-game-demonstrated/#continued
Personally, I think this is not only feasible but also highly desirable. As do all of the people I’ve talked to about it so far: enthusiasm all around, most would like it now.
25/03/2009 at 16:54 Cian says:
With ISPs currently whinging about the demands video streaming places on their services what are the potential consequences of something like this taking off?
25/03/2009 at 17:45 Matt says:
The video epsy’s linked to answers quite a lot.
He talks about the average for 720p being somewhere around 2mbit which is around 240kB so it’s around 4kB a frame, which is mightly impressive.
25/03/2009 at 18:12 adam says:
I am sure that whoever is betting “the farm” on this working has thought of most if not all of these ideas. Latency, numbers of computers needed, etc. Are pretty basic problems that in the 7 years of development I am sure they have thought of. They have had 7 years to fix problems people think that they haven’t come up with solutions to problems you thought of in the last 7 hours??? I for one HOPE it works because I don’t have the $$ for constant computer upgrades.
25/03/2009 at 18:14 Dave L. says:
All the comparisons to the Phantom are completely off base. Phantom didn’t have any publisher support, it had some venture capital and some other independent non-industry financing. Phantom didn’t have any working hardware, it had a mock up of the console. Phantom didn’t have any network infrastructure, or any solid plans to develop it.
OnLive has significant publisher support, working hardware, and already has a datacenter/server farm in Santa Clara, with plans for another on the east coast.
Does this mean that the service WORKS? Not necessarily. But it’s not vapourware, either, and they certainly appear to have a good idea about what it is exactly that they’re doing. And hey! External beta starts in the summer, maybe people could actually try it before dismissing it out of hand as unpossible.
25/03/2009 at 18:49 Jeremy says:
But hey, I’m 17 and have read some wiki pages about network architecture, I know everything and OnLive is impossible.
(disclaimer: I’m not actually 17, but I am being sarcastic)
25/03/2009 at 18:57 Tei says:
If you have money for a fiber internet connection, you have the $400 dollars that will cost you the latest PC. And local will be always more smooth, and better resolution, and better responsivenes, and less DRM than a datacenter 1000 miles away.
25/03/2009 at 19:47 Bah says:
There’s no way this could be profitable, without huge monthly costs, and even if it is, there’s no way this is going to take off.
The userbase is limited to people with decent connections who are in vicinity to Onlive server hubs because of just how much latency would affect the gameplay and the rigs, bandwidth are going to cost a lot. That, and if this takes off i expect to see monthly bandwidth caps to become way more frequent, as ISPs are going to react to the huge amount of bandwidth Onlive is going to use.
25/03/2009 at 20:52 Matt says:
Actually bah, if you’d bothered to take a look at the video espy pointed out. They’ve been in talks with ISP’s, who are happy with what they’re doing. Somebody even questioned about a 150gb monthly cap, and they basically said it’s a non issue. Check out the link.
26/03/2009 at 03:51 Darius K. says:
I was at GDC today, and I checked out their booth at the expo floor. My biggest gripe with their demos is that they were running at what appeared to be ~15fps. I’m not a stickler for frames per second performance, but 15 is pretty ridiculous FOR A DEMO BOOTH AT A SHOW. Which leads me to believe that (a) they don’t know how to put together a demo or (b) they literally can’t get better than that. Both are disturbing to me.
26/03/2009 at 11:09 Aldo says:
Yes and no… what i mean is, the big companies – the ones that make software or hardware that dominates the PC/networking industry – are pretty good at pushing out products that suit their aims above that of consumer. Usually there’ll be peripheral benefits to the customer, of course, but generally most things by MS, Sony, et al are (IMO) developed and sold to further that companies overall ambitions.
What i mean is, companies can still force stuff onto the market and onto people if they’re powerful enough. Call me cynical, but I work (well, for the moment) at one of the worlds biggest hardware companies (being obfuscating here), and they’ll push out unnecessarily expensive or extravagantly pointless products at the same time as slashing benefits, and sacking lots of people. And somehow this works, and people buy overpriced kit (or software).
This tech is seriously just dumb. So many silly things in the way. If you live outside of a major city, this is going to be completely useless. Even more so if you live in a country where bandwidth is limited and you have a set amount of gigs pr month and so forth.
It’s gonna sort of work in the US, and it’ll work in certain parts of asia. Forget northern europe, the countries that were the former soviet union and most of eastern europe.
Part of the US; offhand I think broadband penetration (and availability) in the US is relatively poor compared to much of europe and asia.
26/03/2009 at 13:51 dsmart says:
@ Darius K
Which is EXACTLY what we’ve been talking about here. Apparently they were 50 miles from the station. And running at ~15fps. Now picture what happens when this gets thrust into a real world environment where you have THOUSANDS of people playing.
My guess is that this was always set out to be White box type operation which is only attractive to cable and telco operators – who already have the infrastructure. e.g. you have to be a certain distance from a DSL box to have the service.
Expect Comcast to be the first to sign on to host it. Right after, they’re going to start gauging their subscribers due to that whole bandwidth cap thing.
26/03/2009 at 17:07 Tei says:
Nice article about the topic:
http://worldofdiscourse.wordpress.com/2009/03/25/onlive-terminal-computers-make-their-return-as-gaming-rigs-of-doom-woo/
26/03/2009 at 17:17 Devan says:
The main reason I hope this fails is that is anti-competitive. Assuming it’s 100% successful technologically, people who subscribe to this system would bear the higher cost of OnLive and internet subscription fees mostly due to the perceived savings of not having to upgrade the computer.
However, when people have under-powered computers and are already dishing out monthly payments for “gaming service”, there’s much less incentive to buy competing games which are not offered through OnLive. Developers will have little choice but to develop for the prominent system (currently windows), and will have a much reduced market if they buck the trend.
This could give one or two companies a high level of control over the gaming population at large. Pricing, availability and content would be controlled to maximize profit.
I suppose it would be sort of like if Steam became so popular that it was the only feasible distribution system, and you started having to subscribe to the games instead of purchasing them.
Let them do this to consoles. There’s a reason I stick to PC and relinquishing control of my software is not one of them.
26/03/2009 at 22:24 dsmart says:
…and here is the whole shebang in a more sensible anecdotal manner that I agree with 100%
27/03/2009 at 03:24 Subby says:
Everyone here should please read this article.
http://palgn.com.au/article.php?id=13939
27/03/2009 at 06:01 D says:
Anyone commented on how this new future of gaming will completely ruin user created content and mods?
27/03/2009 at 20:38 Moorkh says:
someone did, but no-one responded… :(
31/03/2009 at 06:06 perilisk says:
Hmmm… for normal rendering techniques, it’s only saving the cost of purchasing a mid-range graphics card, which is not all that much.
However, there are probably techniques (photon mapping?) that are on the one hand easier to run in parallel and on the other hand, have a lot of upfront calculation cost for any scene, with each additional viewer only adding some small overhead. If they get the latency issue worked out, then their approach might actually be really beneficial (and could, especially if it arrives at the end of the console lifespan, give momentum to PC gaming).
21/08/2009 at 11:26 John439 says:
Very nice site! [url=http://aixopey.com/qqaatt/2.html]cheap cialis[/url]