OnLive shuts down streaming games service, sells patents to Sony
arstechnica.com
arstechnica.com
Did anyone expect differently?
Heck, I expect this would present significant challenges even running on a LAN
It's not a matter of technology, it's a matter of physical impossibility.
I can ping google with 5ms latency. That's for the usual ping size packet, allows for 200 round-trips per second.
Do you really expect to push inputs and get streamed video back (and encode and decode said video, which also takes time) with any kind of reasonable response time?
For things like first person shooters it didn't work that well, the latency and input lag was very noticeable. But get that down to a RTT of 20-30ms then I'm sure most casual gamers will hardly notice.
The encoding and decoding is done in hardware and is negligible at best. Also something to keep in mind is that kids today are used to input lag. Both PS4 and Xbox One uses wireless controllers.
The input "lag" with wireless controllers is not in any way comparable to the lag introduced by streaming your game over the internet. For example, tests have shown that my current wireless mouse — the Logitech G602 — has no more latency than a wired mouse. (The range is crap, but that's another issue.) Nor have I had any latency issues using my Logitech F710 gamepad, even with high octane games like Super Meat Boy.
(I've noticed that Bluetooth tends to be significantly more laggy, but that might just be confirmation bias.)
The CPU usage may be negligible, but not the latency. If you live on a place that still has Analog and Digital TV you can flip between both channels and see what's the delay between them. I've seen it being around 0.5s, maybe more, maybe less (Satellite TV is even worse, of course)
And TV's are hardly anything to benchmark against. They are designed to be cheap, and I'm sure most of them do it in software as well, as it's way cheaper. Nobody notices a 300ms latency in the image, as they don't have any inputs to measure it against.
I would say (and I've been saying this all along) is that it's first and foremost a matter of economic impossibility.
Under the assumption that you can make the technology work well enough for a sufficiently large group of potential customers, the business case for streaming games simply doesn't make sense to me. You need to invest huge amounts of money to build, maintain, and update (keep current) highly-specialized server farms that are mostly useless for other purposes, in the hopes that people will pay money to play games on them. All while hardware prices keep going down, and game visuals (including resolution and GPU requirements) keep going up. If you stop upgrading your server farms, within a year or 2 what would previously be top-of-the-line hardware will be so cheap (new in the case of PC's or second-hand in the case of consoles), that unless you are dead set on being able to play the same games no matter your physical location, you can get a much better experience for less money by just buying the hardware.
It has always occurred to me that many people are still convinced cloud gaming is a good idea, but the repeated failures of services like OnLive seem to have proven this wrong for now...
- Playstation Now lets users play PS3 games on their PS4 in a similar fashion (games run remotely)
- The PS4 lets you "remote play" using a PS Vita (your handheld to your console, remotely)
- The Wii U streams content to the controller with remote play (this is different than the PS4 however, it is local only).
- Steam allows you to stream games from one computer to another
Go download Steam on 2 computers, find a free demo or something, and turn on the In-home streaming feature. Press F6 to see the latency chart. Now add your 5ms input lag and a 10ms video serialization lag. The rest of the technology stack is pretty much the same.
With my NVidia graphics card encoding and my Haswell GPU laptop decoding, the lag at home over wifi is only 12ms.
It's definitely not a "physical impossibility" to do this over the internet if your ping is 5ms.
Who the hell has a 5ms ping? Most home connections are DSL or coax, both adding lots of overhead and latency. DSL/VDSL is especially bad. My u-verse ping to the nearest cdn mirror was something like 20-25ms. My work comcast is slightly better with a 10-12ms to a local cdn. When I had residential comcast, I had probably double that. And that's ignoring all the overhead Onlive puts in.
Sadly, onlive would probably clean house in the age of to the home fiber. Maybe 5 or so years from now that service would be more common, at least for gamers. This, of course, means that Onlive would have to deploy servers to every 5-10ms distance. So your typical metro region would be broken down into a few dozen zones, each with its own onlive server, game library, etc. It sounds like a cost nightmare.
I think the onlive guys said that 50-150ms is doable for most games and sold a service that worked well with non-FPS games. The problem is, gamers are FPS obsessed and that level of latency is a killer. That's on top with games being cheaper than ever. A plain-jane PC can get on Steam and buy on-sale games for next to nothing.
Console games typically have a minimum possible input lag of 60ms, and once you add in the controller response time, TV refresh rate, and the design of the game's engine, it's common to have 100ms or higher. For example Destiny on PS4 has a minimum input latency of 100ms ( http://www.displaylag.com/destiny-lag-factor-review/ )
Point is, consistent input latency is much less noticeable than people seem to think. When people complain about lag they're usually complaining about spikey/inconsistent lag. Consistent 150ms is not a killer. There are billion dollar games that have shipped with worse.
For single player games, yeah, you're right, a person can handle 100ms or so, but that's mostly because he isn't playing against anyone else and the game is designed to handle 100-150ms delay. It can handle an extra 100ms or so from Onlive without too many issues. Its not pretty, but it works well enough.
I do agree that spikes and lost packets are the most noticeable issue, but even if you eliminate or minimize those, you still have the coop delay problem.
Also some people are just not that sensitive-to or bothered-by bad quality media that the pickier amongst us would consider non-starters. e.g. there are plenty of people walking around noisy streets enjoying their music via earbuds that don't deserve to be free gift with a Happy Meal.
Of course there are games in this category: those on a mobile phone/Portable systems.
But their sales point is: "leasing games and equipment", however, if I'm doing this I expect good/high quality (like going into a LAN Gaming center)
I agree low quality is acceptable in a multitude of cases, but it doesn't fit with their product.
Many games were not viable on OnLive, some because of latency and some because of fuzzy image quality. It was a good service for the super-casual gamer. But those all play games on their phones now.
If you want < 10ms ping, you need to:
1) Be within 10ms of target datacenter (1,860 miles roughly)
2) rip out the TCP/IP handlers in all of the hardware along the route and replace with highly optimized, workload specific IP handlers (you'll want to do this in the kernel, OR in an ASIC if you're really serious about speed).
3) own all of the fiber along the route.
Basically, if you can eliminate all of the finger pointing and just own the network end to end, you could probably implement this today, it's just that no one is at that scale except for maybe Google (note: I have no idea how Google runs their internal network, but if I had to guess, this is probably what they're doing).
Edit: You can get this now, it's just hard and not worth the cost or effort to do (since no one will bear the cost of building a network like this for gaming, or really for any application I can think of... there just isn't a need for this kind of network performance).
I haven't tried OnLive, but the scenarios are very different so I don't think the comparison makes sense.
Leaving my lack of ability to finish games to one side - I wonder how dangerous the patents are? I imagine Sony's strategy is to limit game streaming to owners of their hardware. If they are aggressive with their IP that could leave the rest of us shut out of an entire delivery channel.
Players who took part in that hybrid plan will still be able to play their
purchased games through Steam, but streaming games purchased through
Cloudlift or the older Playpass subscriptions will no longer be usable
after the end of the month.
You may still be able to finish them.