Slashdot Mirror
OnLive Gaming Service Gets Lukewarm Approval
Vigile writes "When the OnLive cloud-based gaming service was first announced back in March of 2009, it was met with equal parts excitement and controversy. While the idea of playing games on just about any kind of hardware thanks to remote rendering and streaming video was interesting, the larger issue remained of how OnLive planned to solve the latency problem. With the closed beta currently underway, PC Perspective put the OnLive gaming service to the test by comparing the user experiences of the OnLive-based games to the experiences with the same locally installed titles. The end result appears to be that while slower input-dependent games like Burnout: Paradise worked pretty well, games that require a fast twitch-based input scheme like UT3 did not."
Was this really a surprise to anyone who knows anything about the technicalities of time critical mechanisms in games?
The guy logged in using credentials 'borrowed' from an authorised beta tester, from more than twice the recommended distance from the server, acknowledged multiple high latency (due to distance) notifications, and the best he could do is damn the service with faint praise.
With a single word.
"DUH!"
I think the results are as expected or even slightly better than expected (at least from my viewpoint). It shows that something like OnLive will be workable in the future with slightly faster interenet access.
My problems with OnLive are not related to the technical side. Even though i am mostly a casual gamer (at least since i gave up WoW) and i could profit from Pay-per-Hour, i am not sure i would like this. It would require a lot of trust from my side, OnLive has still to earn.
CU, Martin
The menu video seems to be available, but the in game videos now give:
"This video is no longer available due to a copyright claim by OnLive, Inc..."
Are you kidding me? Of all the arcade racing games I've ever played in my life this series ranks up as one higher ones in terms of needing to have fast reflexes (thus, input delay would be a huge factor).
I don't really see this catching on if I'm honest.
People that play graphically intensive games online are likely to also play those games in single player, or other single player games as well, and I doubt OnLive will be serving single player games in this way, and even if they do who is going to want to play a single player game with all the lag of a multiplayer? And most of these people will have higher-than-average spec computers anyway, so playing games wont be an issue.
And games like WOW etc. aren't particularly demanding of hardware, any mid spec or even most low spec computers made in the last few years will be able to handle it no problem.
And what about people that want to play games over LAN sometimes? Having the game installed locally on your machine is much better than having it stored miles away on someone elses server that you don't have any real access to.
I am a leaf on the wind, watch how I soar.
What is it with all this 'cloud' stuff?
I've got half a terabyte of storage, a pretty good graphics card with shader support and a nippy CPU.
When there are raytracing cards with inbuilt physics, I'll enjoy a slightly more realistic gaming experience on my local machine, thanks.
Until then, I'll have to go with pretty realistic and the only significant cause of latency being my old neurons.
GOML.
Read: "excitement (from clueless arts majors masquerading as tech journalists) and hilarity (from anyone with even a remote shred of knowledge of the technologies involved)".
Look, this tech may - may - be workable for SimWarConquer, but for anything that's reaction based? No. Not going to happen. There is no technobabble solution to latency, and anyone who tells you otherwise wants your credit card number.
If you were blocking sigs, you wouldn't have to read this.
I came here expecting to see a belated "First!" post followed by a joke about lag.
Based on the posts in this topic and how many perfectly sensible posts are mod'd down it seems pretty obvious to me someone with some kind of interest in OnLive is moderating.
First time I've seen such distinct bias in moderation (both promotion of posts and demotion)
Look out lads, metamoderatings gonna getcha!
Mod parent up. If you can't keep up with the upgrade cycle required to play the latest PC games, buy a console or play older games.
What, you want to have your cake and eat it too?? Have you not learnt anything from the recession??
I must admit, I've not actually played it, but if it's anything like the other Burnout games, millisecond reaction times are kind of important. It may be that he has having a hard time picking up on la instinctively because of the analogue controls but I doubt the reaction time increase would stand up in serious play.
No kidding!!! What do you say at this point?
Even if the fps's didn't play as well - the video of the game playing looks really impressive. I can understand the games playing better with a controller it's going to be smaller, more consistent movements which I imagine would be a lot easier to track.
Is this going to appeal to the hardcore, twitchy fingered, give me super graphics or I'll puke gamer? Nope; but for the rest of us it looks pretty impressive.
I for one am looking forward to seeing what they do with this.
While I've been mildly interested in OnLive, my biggest excitement over this was a confirmation that a streamed remote desktop session with real good responsiveness (say LAN) could be had soon. I even started poking around for similar systems that actually streamed the desktop in 2mbps or similar video stream with interactivity, but alas, it seems like no one is working on this issue.
So, I'm open to suggestions, is there any current existing remote desktop server/client system that actually streams the desktop in the OnLive fashion, or is anyone working on one similar to this? (And I do not mean in the old VNC fashion). I believe such a system is very feasible. Imagine being able to stream your desktop onto thin/mobile devices just like you were on it, being able to play video (at least) would be so much better than the current remote desktop offerings.
In a nutshell, I want this applied to remotely stream desktops with full control ala VNC but similar to OnLive.
I recently bought a PS3 with some games. When I started it I was welcomed with "You need to install the latest PS3 firmware now!". So I had to wait for it to install and reboot. Then I inserted a game and wanted to play, but I was welcomed with "Updates have been found for this game and need to be installed". Which is pretty much identical to the PC, but there you often have the option to install the patch.
I told you so.
http://slashdot.org/comments.pl?sid=1492974&cid=30591584
http://www.pcper.com/article.php?aid=859
http://www.pcper.com/images/reviews/859/bw.jpg
-Woof woof woof!
Reading the original article it seems that using an Xbox controller is much less sensitive to lag than mouse and keyboard. Presumably this is because the input device itself is less precise and less responsive. I love mouse and keyboard control though so I really really hope this doesn't catch on and start a move away from good old mouse-look.
If you can't keep up with the upgrade cycle required to play the latest PC games, buy a console or play older games.
The problem with playing older games is that either the matchmaking servers end up switched off (e.g. DNAS Error -103 on PS2 games), or if not, the established players tend not to be friendly toward newbies (e.g. "gtfo n00b" on several classic first-person shooters).
The end result appears to be that while slower input-dependent games like Burnout: Paradise worked pretty well, games that require a fast twitch-based input scheme like UT3 did not."
It was going to be pretty damn obvious that this would be the case.
Unless they invented some sort of tachyonic backwards compatible networking layer, it wouldn't have worked.
The only other potentially manageable solution would be servers at exchanges, but that will cost them their entire body several times over.
It still won't make things *that* pretty though.
I predict that cloud gaming services will utterly dominate all gaming. Within 5-10 years, virtually all new titles will be released exclusively for cloud gaming services, and will not be available at all otherwise. Consoles as we know them will become totally extinct : the next generation of consoles will be the last.
The reason is economics. It isn't hard to see the enormous forces that will push cloud gaming to domination.
1. All game publishers will be paid for every hour of game played. This increases their income, which will increase the supply of high quality games. This is because with cloud gaming, piracy is basically impossible, yet the forces that drive pirates are also mostly eliminated. You won't have to plunk down $60 to try out a game legally : you'll be able to play any game for $1 an hour or $40 or so a month.
2. Instead of there being 3 console platforms + PC, there will be just one platform : the PCs in the cloud gaming data centers. Most likely, the cloud gaming providers will soon release development kits that are a PC with the exact same hardware and OS image as they have running in their data centers. PCs are already the easiest platform to get a game running on, with the best dev tools. Now game publishers will be able to develop their games exclusively for this platform, with a fixed hardware base.
3. The overall costs of gaming will be lower. Instead of every gamer needing their own CPU + GPU, whether that be in a console box or a gaming PC, they'll just rent a portion of one from a central service. Even after collecting profits, cloud gaming providers will be able to provide gamers with gaming services for much annual lower costs than buying a gaming system and games.
4. The big criticisms in this article will be eliminated once there are more cloud gaming data centers, located all over the United States to put one close in terms of network switches to every ISP customer. (reducing latency to 30ms or so, enough to eliminate perceived lag) Also, there are some tricks with the mouse input that could eliminate the overshoot described in the article. (sync the mouse coordinates on the client with the server) The other big critism : inferior graphics quality : will be mostly eliminated with more bandwidth dedicated to the video stream. The author notes that the current beta client uses only about 1 megabit for the video. 5 megabits would greatly increase the image quality.
5. Next generation games that blow your mind graphically will now be practical. Right now, you can't develop and sell a game that requires cutting edge hardware to give photorealistic graphics. Consoles are years behind, and you can't write a PC game to require a new $2000 PC that only a few gamers have at any given time. With cloud gaming, however, that will be entirely practical : if you're willing to pay a little more per hour, you'll be able to enjoy Crysis 4 maxxed out with smooth as glass, uber realistic graphics.
6. While one form of lag is introduced with cloud gaming, it eliminates another. Since each game client is running in a data center with excellent internet connectivity, latency BETWEEN clients in a multiplayer game will be virtually eliminated.
7. The tech support nightmare of supporting games due to hardware problems is mostly eliminated. Games will also load far, far faster because the cloud gaming service can just switch your session to a PC that has already loaded the game.
8. Eventually, enough hardware to support online will be integrated into new TVs and blu-ray players, so basically anyone with a TV and a spare USB mouse/keyboard or USB gamepad will be able to enjoy PC games in their full glory.
9. The resurgence of mouse/keyboard using gamers will mean that PC game genres like RTS will make a big comeback.
And lots more reasons.
The reasons slowing down cloud gaming?
1. ISPs have to have a contract with the cloud gaming provider, and to use QoS to
... because I couldn't even stream the videos without jitter. :)
There was even more latency than you expected, you insensitive clod!
Do what thou wilt shall be the whole of the Law
The major problem isn't overall latency, it's little spikes of latency on an otherwise good line. A moment of 100ms lag on an otherwise good line doesn't matter for online games because of client prediction and at worst it's a tiny moment where the controls don't seem responsive. It's not a problem for normal video because they can buffer 250ms or 500ms or 1000ms of video without any problem. But on this they can't do any significant buffering or the latency will be too much to play.And even 100ms of sudden latency will cause the picture to lag or freeze or jump. It might only happen occasionally but I suspect people won't put up with it. And they can't do anything about it either, even if your ISP is only 10% loaded on its lines and routers, there will be times when all that 10% send packets at the same moment and they get queued in a router somewhere, just for a tiny time but tiny little amounts of jitter like this are normal and expected and to be honest I think will be the downfall of this project because there is no real way to deal with them. But I guess we'll see :)
My wild guess is they'll put the servers where the people are
Not everybody lives in New York or Los Angeles. There are 200,000 people in Fort Wayne, Indiana, including myself. What are the odds that I'll get coverage?
I dunno. I had never even heard of this service before but... I am a gamer but I don't play that much FPS games. Mostly RTS, RPG, etc... In those, latency really isn't nearly that big of a problem. So, if I could play Dragon Age on a laptop with good graphics and without worrying about the damn thing crashing when rendering a highly detailed cutscene... Great!
There are still large areas of North America stuck with either stone-age Dial-Up (in 20-freakin'-10) or slow expensive satellite. Like mine (I cry myself to sleep over my 1200ms latency) This is absolutely a no-go there. Obviously.
Now, in better places, I'm sort of out of the loop. Whenever I've spent time in cities, either visiting my brother in Ottawa or living in London (Ontario, not the good one) for a few months at a time, it's been my experience that even connections that are supposed to get up to 1MB/sec would be lucky to get that in practice, especially at peak times. Furthermore, the sheer amount of lagspikes, connection hiccups, or general time when the interrnet craps out for no apparent reason makes it seem like you'd be dealing with one frustration after another. The number of times I see people get DC'd on World of Warcraft seems to back up my theory that staying connected, and maintaining a constant connection at 5KB/s or so (for WoW) is difficult enough, doing the same for a (whopping?) 1/MB/s while keeping latency under 100ms would be hellish.
So is my experience with the Internet indicative of the general population, or have I just had the misfortune of having terrible service? Can people really keep 1MB/s sustained, without lag hiccups, DCs, lost packets, etc, while keeping under 100ms?
And the problem with consoles is that they're not as good as PCs. I've checked comparative screenshots and the PS3 can come close, but the XBox360 tends to look shoddy. PCs these days can pump out way more frames at higher resolutions than consoles (actually, I think they've always done that, but consoles are at least up at decent resolutions now), so I'd still rather have my PC.
Check out VMWare's PC over IP protocol and its implementations.
I get great interactivity over high latency and low bandwidth links. But then again, I use this for regular desktop stuff and sometimes watching videos (mostly flash), not hardcore gaming.
IMO, PCoIP is the best there is at the moment.
This is an obvious pump and dump scheme, unless they have somehow unlocked technology previously unseen and unknown by mankind, and have done so for the purpose of playing video games.
Because guess what? In the real world, people live all over. Onlive isn't going to be able to say "Just move closer to one of our data centers," at least not if they want to pitch themselves as the "cheaper than buying a graphics card" option. Sounds to me like they've been controlling who gets in to the beta to try and create an overly rosy impression. This guy was a more realistic test, a person who doesn't happen to be near their few locations.
That's just the reality of this. If it is to work well I can't only work well for a few people in a few locations.
Also the more revealing part was just how bad things look, just how much the compression degrades the image quality. The difference between the local and remote screenshots is almost the difference between SD and HD. While the Onlive stuff is 720p, technically, a ton of the detail is getting lost. That's just what you are going to get when you try and jam HD video in to a 1mbps stream. Only problem is that detracts from one of the supposed reasons to get the service. The lower the resolution and image quality, the lower end graphics card that could handle it on a local system. So Onlive isn't giving you the same experience as a $400-$600 graphics card, it is giving you maybe the same experience as a $50-100 graphics card. Well then, makes it much less worth while.
So initial results seem to show that the doubters were right:
1) Latency will be an issue. If you don't happen to live near their datacenters, your latency may make playing difficult to impossible.
2) Quality will suffer. They don't have some magic voodoo compression that makes everything look perfect, their compression is like everyone else's and trying to do 720p @ 60fps equals a good deal of detail loss.
3) Even if you have a good net connection, if there are problems or congestion, the service will be unusable, meaning you can't play your games whenever you want.
Makes it not so attractive as they hyped it to be, especially against powerful $100 graphics cards (the low-mid range of graphics is great these days) and $200 game consoles.
What you have to remember about ping is it is more or less testing minimum time. The payload of an ICMP packet is very low. With video data like this, you have more payload. So you not only have to count the transit time from the datacenter, but also how long said amount of data will take to transfer at your line speed.
For example, say each video frame is roughly 50kbytes. If you had a line that was only 50kbps, well then it would take a full second for you to receive a frame, even if the latency on that line was 1 ms. You'd start to receive it almost immediately, however it'd take a full second to get it all. Now if you had a 5000kbps line, you'd receive the data in 10ms. Much better.
So a service like this depends on both speed of the connection and latency. You'll note that he said it didn't work when the speed of his connection was around 1.5mbps. Now at first glance that should be fine, 1mbps video stream, still plenty of overhead. The problem is that the time it would take to transfer the video data at that point would add too much latency.
Thus to work well with this you need high bandwidth and low ping to their servers.
Just how many remotely rendered fps are you going to get, even just down the hall from the server?
What would I use remote rendering for?
Maybe if I were making the next Pixar movie and the laptop wasn't up to it, or I wanted to render huge datasets from MRI scans.
Gaming? Heroic idea but as retarded as a brainstem in a blender.
I did this back in 1996. I was wardialing, and found somebody's open PcAnywhere connection. So I connected to it and attempted to play Solitaire over 33.6 dialup. Needless to say, it was a pain dragging those cards around.
85ms is not a high ping at all. If Onlive considers that to be outside of their acceptable range, they've got a nasty surprise coming when they try to open it to the public. Lot of people going to have a ping of that or higher.
I've got a reasonable good connection here, business class 12/1.5 mbps cable. Represents a mid to high end home connection. Minimum latency on the connection is in the realm of 25ms. That is about what it takes to get out past the CMTS and so on. For good sites, I'm usually in the 40-50ms range, provided my ISP has peering to their ISP and they are in a relatively nearby state. However, I easily see pings in the 80-100ms range to many locations, including a number of TF2 servers I like. 150ms+ isn't out of the question either, I've seen that to various places in the US that are a bit distant that don't have a very direct connection.
So even with an over all low latency connection, you can easily find it up in the 80+ range. Gets worse for ones that aren't as good. When I had DSL, minimum ping was more like 50-60ms. It took much longer to get to the DSLAM and out than the cable connection does. Also peering wasn't as good so it was rare for me to find a site with a ping less than 100ms.
This is just the reality of broadband. You don't get nice low latency all the time everywhere. It is slowly getting better, and I'm given to understand that FIOS is even better latency wise, but you can't look at the best and say "Oh well that'll work great!" You have to look at what most people will be dealing with, especially most people who won't spend a ton of money on Internet. After all if your business model is "This is so much cheaper than buying high end hardware," you can't very well then say "But you have to buy high end Internet."
Only happened to me twice. You just choose I nice, friendly server and stick with it.
Dilbert RSS feed
How will they get user maps and mods on this?
if all gameing where to go this way that will kill that.
also how about free games / small game developers that may be shut out of this?
He says that this may be in cable boxes some day but with with the VOD control lag that you see now days that may not work out that well and cable co's don't have a lot of bandwidth for this some don't even have room for all people who want to use VOD at some times.
caped / metered ISP's will kill this 1meg /s min is to high for many caped plans. Even comcrap 250gb will limit this.
My point is that it can be hard to find a nice, friendly server for an older game, or at least one where you survive in a round long enough to learn anything.
any one have any more info on it? any one tested it?
http://www.israel21c.org/culture/casting-a-spell-on-cable
http://www.indiantelevision.com/headlines/y2k9/july/july267.php
Until we see a 10-fold improvement in average bandwidth and latency, a service like this will have limited appeal.
Personally I vote for improviong the network, but you know that just me ;-)
True, but if you make something available only in major cities, you'll end up with residents of minor cities whining that they can't get the service, and that it shouldn't be considered "in production" until it reaches 85 percent* of the population. The whining can be even more effective when performed by a competitor; Verizon's ads blasting AT&T for its spotty 3G coverage have done a good job of creating an impression that "there's a map for that".
* The U.S. DTV transition was originally supposed to trigger analog switchoff at 85 percent coverage before it was changed to a specific date.
My assumption has always been that their business model would be to sell the kit to ISPS who would install the servers at local exchanges and charge for gaming as a premium service. That would seem to solve most of the problems. The physics only really becomes a problem when you assume central servers away from subscribers
If I've been paying attention to the hardware storage roadmaps isn't the next "DRM" idea that the games will be stored in a very large uncompressed format in order to make it costly to copy and pass around the games?
Isn't this movement going to work exactly contrary to this streaming cloud gaming concept?
I suppose they could make a stripped down compressed version, but another hardware medium has been doing that for years. It's called cable. Take a look at some of your channels and see how compressed video data looks. Probably only a matter of time before they would have to do something similar to provide this kind of service moving forward.
Maybe this kind of thing will finally give Comcast enough legitimate reasons that customers can and do exceed the 250gig/month limit.
So latency is always going to be an issue for twitch-based games. But I tell you, sitting at work right now, I'd rather be doing nothing for my lunch break other than playing Dragon Age: Origins. But I can't very well install it on my work computer, but with a service like OnLive, I could log on and play.
I was at the GDC 2009 where this was announced and I immediately thought it would be dead on arrival. I still feel that way.
Doubters should really watch the Columbia University presentation. It's entertaining and very technical and will probably address your every concern. Too many genuine experts here don't know what they are talking about because they are ignorant of the way OnLive actually works. It's more clever than you probably think.
YouTube Mirror
http://www.youtube.com/watch?v=2FtJzct8UK0
Original
http://tv.seas.columbia.edu/videos/545/60/79
+0 Meh
I've been playing Classic DooM with http://www.skulltag.com/ and they've been fine for me, so the hostility isn't with all older games!
Is it available in Nebraska?
who gives a shit if OnLive loses business in areas they haven't even set up data centers?
Anyone who understands fixed costs. I doubt OnLive's ability to make a profit from a fraction of a fraction of a fraction of the population. Fraction 1: gamers. Fraction 2: those who live in range. Fraction 3: those who choose this service over other available services, such as Blockbuster or free Flash games. But if 47 out of 50 U.S. states will be fully covered at or soon after launch, as someone else claimed, I have no complaints.