Richard Leadbetter writing for EuroGamer, on why the streaming OnLive game service is so unlikely technologically that it must be practically impossible:
1. OnLive has mastered video compression that outstrips the best that current technologies can achieve by a vast margin. In short, it has outsmarted the smartest compressionists in the world, and not only that, it’s doing it in real-time.2. OnLive’s unparalleled grasp of psychophysics means that it has all but eliminated the concept of IP lag during its seven years of “stealth development”, succeeding where the best minds in the business have only met with limited success.
3. OnLive has developed a range of affordable PC-compatible super-computers and hardware video encoders that are generations beyond anything on the market at the moment.
That article pretty much echoes my opinion of the system. A good idea for LANs, I could imagine, but the minute your using public pipes and travelling long distance through multiple switching sites, it will be unplayable.
Unless they do have these magic codecs/anti-lag systems, which sees unlikely.
It *might* be sort of playable over on Korea’s awesome internet or Japans for that matter.
http://www.worldpoliticsreview.com/Images/commentarynews/broadbandspeedchart.jpg
Unless they stick to simpler games I don’t think it would be all that playable even in the best of times.
Now flip this on its head and ask what types of games could this be good for? Turn based games, d&d – mudish games with lots of chat.
Hello kitty island adventure, neopets 3D etc I think could be interesting.
Also, heavier-than-air flight will never be possible.
Obviously they don’t have magic cheap supercomputing. However, the video encoding load is fully subject to Moore’s law, what’s impossible now will be cheap in short order. Plus, the bulk of work in video compression has been in offline rendering, not real time, so it shouldn’t be a surprise if OnLive’s system pushes the state of the art. Latency on consumer internet is also likely to drop over time, and they can spread their datacenters out towards their customers to somewhat improve this. OnLive will be viable eventually with enough funding, if it isn’t now.
The most interesting question that this article raises is how they’ll handle peak loads. Other major cloud providers like Amazon and Google have obvious ways that they can use excess machines, so spares kept around for breathing room won’t go to waste. Will OnLive find a use for a metric fuckton of video powerhouses, or will they deny service to would-be customers when everyone wants to play at once?
If your compression algorithm is highly parallelizable and your hardware is designed only to handle one resolution, you could quite possibly get that sort of low-latency decompression. However, on commodity CPUs like on every desktop computer today, you won’t get that.
And what will they do with the machine capacity in low-usage periods? Because the machines have to be close to the users, you can’t have gamers in Korea or Germany making use of servers in California while America sleeps. Will the system be designed to be usable for less latency-sensitive tasks during slack periods? I imagine they’d have a lot of spare CPU cycles they could rent out.
Trying to scam investors I see.
The only way I can imagine this working is that they cheat in every way possible. All media compression is based on cheating the senses, but this demands marketing level cheating.
If you are defining your own compression systems, there is no limit on what you can get away with in marketing talk. For instance, nowhere is it said how a video frame is composed. Send 5 fps 640×320 stream to client and up convert to 120 fps 1950×1080
if you like. Obviously that looks like shit, but you won’t show that to investors.
Yeah, I came to much the same conclusion. It’s unworkable for business as well as technological reasons.
http://lpar.ath0.com/2009/03/25/is-onlive-the-future-of-the-video-game-industry/
Is the “streaming game service” the “perpetual motion machine” of the gaming world? I never did see anything about this new service that *didn’t* make me think of the Phantom. (So, when should we expect those at retail, again?)
While it probably won’t be the amazing paradigm shift in gaming that it’s marketing itself to be, there’s got to be something to it. It was demoed live at GDC from a server approximately 60 miles away, and it has several big-name game publishing partners. If this was complete bullshit, I doubt you’d see giants like EA and Ubisoft giving this public support.
Every.
Single.
Network.
Hop.
Adds.
Lag.
If you have a unixy computer, you can type traceroute google.com and see what the latency is.
In my case, that amounts to an average roundtrip of 68 milliseconds, with a standard deviation of 44 milliseconds. In other words, from me to google and back will vary between 68-44 to 68+44 milliseconds.
So that’s a “frame rate” of that WILL vary between 9 frames per second to 44 frames per second. That’s not acceptable, right there.
But the other 99 nails in the coffin — this 9 to 44 fps assumes that the time it will take them to decode the input data you send them, calculate the results, render it, and compress it is ZERO.
Let’s be really generous here, and assume that they have super-genius programmers who can predict your typical game input, and they can pre-render the results so that as long as you play along with their predictions, they actually can calculate far enough in advance to make that “net” time REALLY CLOSE to zero.
What happens whenever you jump sideways or yank on the steering wheel? The game takes ONE TO TWO SECONDS to react AND the graphics go to shit.
How much will you pay for that? I didn’t think so.
I can’t even stand the delay added by playing standard internet connected multiplayer games over a wireless network. I can’t imagine if all the graphics had to be sent over the ‘net.