Meet games-streaming Stadia, yet another thing Google will axe in two years Challenging Amazon and Microsoft for the attention of gamers and for the billing of game industry customers, Google announced a planned game streaming service called Stadia at the Game Developers Conference in San Francisco on Tuesday. Stadia, which will be integrated with YouTube, is expected to launch toward the end of the …
A few bucks a month? You're having a bubble mate, this isn't going to be the app store.
You'll get access for £15-£20, month not including 'prime ' and new releases. Oh and now you don't need the box under telly, a new release will be £70+ anyway. Micro transactions on top of course.
Yeah I'll pass thanks.
"So who is going to pay to dig up all the local streets to lay fibre so that there is enough bandwidth for this?"
Forward-thinking places have conduits that run under the street. Then it's easy to lay new stuff. Unfortunately, Britain has the problem that it was built too soon, before this stuff became obvious. There are several roads in London, for example, that are designated as 'full', and cannot be accessed for new works. With sewers, phone cable, water pipes, electricity, gas mains, broadband, and other things, there is no more room under the street, even if it is dug up.
Research is happening at the moment to develop sensitive ground-penetrating radar that can map this network of pipes. Because nobody bothered to record where this stuff was when it was laid.
How are they keeping it on their backbone all the way to the client? Are they planning on putting in fibre to every POP in the world? or are they only going to be providing this in a limited number of locations (where they have deployed their own fibre) and then just close it down when it doesn't have enough customers?
I mean it's not hard to imagine Google happily ripping out your spine, if they thought it'd sell ads.
Reality is it'll be another VPN overlay like Sony and MS's networks. Which makes them a convenient walled garden that keeps out viruses and malware. And indies or competitors. ISPs may be lucky to get invited to install racks and capacity in their datacentres to stop gamers blaming the ISP when they can't open a lootbox.. Or find laaaag.
Which also means the US (sorry Comcast customers) won't get completely 4K'd over by Google. Or they will, but they'll blame Comcast for not investing in their network. I mean Google managed to build out gigabit fibre to the masses..
On the plus side, it'd probably be tax efficient to bung millions at developers to tempt them into this new platform (that may or may not be exorcised at Alphabet's discretion) given the billions Alphabet has in it's various slush funds.
But then if its just a VPN it's not keeping the client on their low latency backbone. It's using other providers backbones therefore subject to that providers latency and QOS.
So if they are not planning on putting fibre everywhere they will need boxes at every isp, but still will be subject to the isp backhaul from the POPs
It's using other providers backbones therefore subject to that providers latency and QOS.
Therein lies a potential political problem. You're right that unless Google builds out it's own access network, then it's dependent on creating a virtual network.. And it's going to be 'Internet' to all intents and purposes. So to have a decent gaming network, you'd want to be able to prioritise traffic, manage buffering and queueing to prevent packet loss or lag. So like you say, implement QoS.. Or persuade ISP partners to implement QoS.
Snag with that is Google's spent millions lobbying for 'Net Neutrality', so rather awkward to change tune and start shaping packets instead of legislation. Which is perhaps why Google's going to pretend it'll be using it's own 'backbone', so then it's a private network and QoS/CoS is OK. It may be able to mitigate some challenges by trying to persuade ISPs to add dedicated peering connections tor the gaming network, but those ISPs may point out that it's a commercial service, and they should pay for capacity instead.
I have a colleague who rents a powerful VM out at about £30 pm so he can play games in good graphics. Otherwise he would have to buy a gaming pc which would be out of date soon. Supposedly the graphics are amazing and justifies the cost of the service and requiring a good fibre service from the ISP.
Not something for me because I wouldn't even have the time to start a game without the missus or the baby needing something.
Fantastic, let's take the greatest disadvantage of consoles, which is a fixed architecture that rapidly becomes obsolescent, and replicate it a millionfold in the cloud where you have to use a very long piece of wet string to communicate with it. Only a company the size of Google could come up with such a stupid idea.
> Only a company the size of Google could come up with such a stupid idea.
Eh? It's the same idea as has previously been implemented separately by OnLive, Sony and nVidea. The idea is sound, but there were technical hurdles preventing a great user experience, technical hurdles that Google seems confident it can mitigate.
Google's architecture isn't fixed like a games console - their system is designed so that multiple GPUs can be used to render a single instance of a game, allowing, for example, someone to pay a bit more for greater fidelity graphics.
It's worth noting that a lot of console games today are dependant upon 'a long piece of wet string' in that they are multiplayer across the internet, typically with one player's console acting as a host to coordinate the game. For this reason it may be that there's less lag (with regards Bob's bullet hitting Alice's helmet) in a streamed game than in a traditional console online shooter.
The presentation last night was a little thin on detail so I couldn't get a handle on how any virtualisation would work, but the idea that you can use multiple GPUs per instance rather than a single more powerful virtualised GPU suggested to me that any virtualisation was limited because it would surely add to the latency. So you are kind of stuck with the hardware, which having seen the way that a single Stadia GPU struggles with fluid dynamics, is going to make multiple GPUs pretty essential. I mean, saying "hey, look at the improvement with two GPUs" is pretty much an admission that a single one isn't up to it.
The real elephant in the room though is raytracing. I know that you can ray trace in realtime without the Nvidia RT hardware because Nvidia have had to shoot themselves in the (elephant's) foot by porting their system to earlier GPU platforms due to the slow take-up of the RT-enabled ones, but if it does take off in a big way then Stadia may just be end up as an expensive liability.
Normal multiplayer games don't have such a large amount of data sent down the wire.
Streaming, there will always be stream-delay - the time taken to encode, send and decode the video is always going to be additional to whatever you have in your local hardware loop.
Fine for some games maybe, but useless for anything that requires quick reactions, where fractions of seconds count.
There is a reason OnLive has gone the way of the Dodo, and I don't actually know anyone using the Sony or NVidia cloud solutions. They are not massively popular.
I know one person who uses NVidia's in-home streaming, and I and a couple of others have played with Steam-link.
Even purely on a local, wired, gigabit network, the additional lag is noticeable - and that's before you leave the LAN.
The big problem of multilayer gaming isn't capacity but latency, and it's always going to be at least as slow as the slowest player relative to the server. This idea isn't such hitch if the players are relatively remote to each other (as your biggest lag will be syncing the clients), and if you're a specifically local clan or a LAN party, then This Is Not For You.
"the greatest disadvantage of consoles"
Not really, how can the same hardware across the board by a disadvantage? And as for it becoming obsolete, I would call it aging gracefully. My Atari 2600, SNES, N64 etc get a look in more often than not nowadays and seem to be increasing in value according to ebay.
I like tangible hardware but if they can tackle the connection fears then I'm quite excited for this.
Pay for games & hardware that might get dropped at any time?
Pay for games that might get yanked off the service or censored at any time?
Sorry but no, a NetFlix style service for games and a really bad idea IMHO.
If I can't buy physical media or download the installer to run on my hardware I'm not interested.
Many games are only as good as there are other players to play against and servers to run the games. It tends to be that most of the people who played Halo 3 switched over to Halo 4 when it was released, leaving the Halo 3 servers a bit devoid of players. So, that's the 50 quid for Halo 3 disc, plus the ten quid a month for the Xbox live Gold subscription... for a game that itself gets superceded after a year or two... This staus quo doesn't sound much better for the player than just paying a monthly subscription.
But then what about the other games that are true masterpieces, yet are devoid of any multiplayer mode (or it being secondary in that the game is still fully enjoyable alone)?
I know those are less and less prominent than before due to PvP and Battle Royale being the 'thing' right now but they are still out there.
We've had around 30 years of console games and PC games coexisting, with their respective strengths, genres and cost of entry. At no point has there been any threat of gamepad-based consoles dying out, and people are still spending big money on gaming PCs.
Classic single player games will always be available.
"Many games are only as good as there are other players to play against and servers to run the games."
But at least equally many games don't involve or require online multiplayer. Of those, if we're talking about modern games, far too many artificially require the use of a server somewhere in order to allow the game publisher to maximize revenue, but lots don't.
You won't be paying for games, you'll be paying a subscription fee to access games. It will be a monthly fee. If they vanish, you stop paying.
By your logic Netflix, amazon prime et all will be massive flops. After all, who would pay for access to media they don't own? Right? Why is it a bad idea fro games but good for other media? Kindle unlimited, the various music services. All work and are flourishing off this very principle.
It'll be a hit. If it works, i'll be a very happy chap.
Video gets buffered. Games need to be low latency.
You typically don't watch a movie or series more than once. You watch it, you move on to the next thing.
Age of Empires II, a game released in the 90s has an active playerbase and competitive scene. It's an extreme example, but there are reasons to keep playing games that you have already played before. PSNow is ~£20 a month I believe. A game like Forza Horizon 4 is on the higher end of price, at ~£60. If you like racing games, you may very well want to play it for more than 3 months. If so, it's going to cost you more.
There are a number of reasons.
"You won't be paying for games, you'll be paying a subscription fee to access games. It will be a monthly fee."
Yes, and that's one of the primary things that make this whole approach a nonstarter for me. I love going back and playing games that I bought decades ago, and it would be a serious loss if I couldn't do the same with games I buy now.
That's why I don't buy games that require phoning home in any way in order to install or function. Which is why I've stopped buying games that aren't available on GOG.
I don't think so. Outside of a relatively small number of collectors, I don't remember anyone actually wanting to own movies back in the day. For the most part, most people only watch a movie once, after all. They just wanted to watch them on demand for a reasonable price, and owning a copy was the only way to do that. Things like Netflix offered a better way to meet that need -- even back when Netflix only did DVD rental by mail and had no streaming service.
I think games are very different in this respect.
indicates that watching *other* people play games on you tube has replaced actually playing them.
So I'm not sure why Google is worrying about pandering to the game playing minority on the same site.
Maybe the numbers of streamers are dropping, and the controller will be cunningly designed so that you cant avoid accidentally pushing the "stream my game" button?
Any human-v-human game where some think cheating is ok, whereas the others do not is probably not going to match the definition of "fun" very well for the non-cheating players.
Anyone who likes game cheating is IMO welcome to do so in contexts where that behaviour is an expected feature. If that's what you like, go for it, I don't care. I can sort of see the "backstabbing-the-backstabbers" attraction, even if it's not particularly for me.
But instead, if a "cheating is fun" attitude is based on (unfairly) screwing over players trying to play an honest by-the-rules game - what should we think of that?
Whatever the answer, I don't think it has much to do with whether it occurs in a traditional game or somewhere else.
Back in the day, like in that case, '93, some games included cheat options. Ok they were mostly single player. Fast forward, and many single player games now include multi-player, and no cheat options because that may impact the sale of loot boxes.
Or other players fun, ie people who use aimbots, wallhacks etc to give themselves an unfair advantage in competitive games, which is not fun for players on the receiving end. And if left unchecked, leads to players deserting the game due to rampant abuse and the game slowly dying.
And then there's modding. Way back, in World of Tanks, it used to be possible to extract the meshes for tanks and create your own skins. Which was fun, and also allowed players to create better looking vehicles. Change was client-only, so other players didn't see the results. Then players realised they could also make tanks with neon skins so they were easier to spot, or highlight weak spots so they were easier to hit. So that got banned. And seeing as customisation is a common way to fleece customers via microtransactions, the only way to get a gold AK is to pay .99c
Or there are games like Space Engineers. The vanilla experience is fun, but a pretty empty sandbox. But there's an active and creative modding community that expands the game. More planets? Check the workshop. More ships, objects, encounters, furniture.. all available from the workshop. It's also multi-player and supports PvP, but has admin (aka 'cheat') options to limit abuse.
Or there are triple-A titles that are strangely empty, with voids that can be filled by future DLC. See EA/Paradox for more info. Unless it's Mass Effect, where poor reviews lead to obvious DLC holes remaining unfilled. TL;DR though is it all depends on the game. Kludging multi-player into a single player experience often results in a bad game, and locking it down stops the community making their own fixes, or expanding game play/longevity.
I have no doubt that they can throw down 4K at a pretty impressive frame rate using what we used to call powerful servers with lots of GPUs in a data center, but now must call cloud.
The real question for most gaming is how long it takes for a player action to be noticed by the game, and for this you are likely north of 5-10ms because physics.
A lot of games are already human Vs human, so the question isn't how long it takes the game to notice the players' actions, but how long it takes a virtual bullet fired by Brighton Bob to register as hitting Cardiff Charlie's helmet. This is a solved problem.
Indeed, the latency will probably be lower if the game is hosted by Google and not by Dundee Dan or one of the other players as is common today.
Latency latency latency!
Ever RDPd across the Internet? Even "playing" Excel is often intensely irritating.
I can only assume nobody in Google has ever played a first person game...
At 60fps, the latency from mouse to visual is 16ms or 32ms depending on the game.
My ping to Manchester is currently 28ms, and to Dublin (nearest Google datacenter) around 40ms. For reference, "slow" response starts being noticeable to most people at around 30ms.
So at 60fps, the game I see takes at least 3 frames just to get a mouse move from me to them and back, so in reality what I see is no less than 5 frames behind my input assuming the server and my browser can both draw it in less than ~10ms each.
So that "60fps" is 83ms latency, roughly similar to 12fps (albeit smoothed with really good interframes)
That's the best possible case, it'll be far worse for most people most of the time.
First-person shooter games would be unplayable against anyone with mediocre local processing, perhaps at all, and VR would be literally nauseating for absolutely everyone.
As a video streaming service it certainly could have legs. Twitch is often rather annoying, so making a better UX would be very valuable.
As a remote gaming service?
Not unless they install the servers in your local cabinet.
That's all the Googlers actually try these things out with... the 1TB+ connection in their office with 50 inch monitors or the very latest smartphone... that IS the real world to them. They even get taken home in a Google bus with wifi. They are aliens in their own city. Most of them probably have no idea what latency is as they never experience it.
This post has been deleted by its author
That sounds a lot like many modern websites too. Developers test them on their own ultra-fast connection with very powerful computers, but never bother to test it out on a weak computer with an slow connection. Even on my 400 meg internet with a Ryzen 7 and 16 GB of memory some websites bog down my computer when I turn off NoScript.
Google needs to test this with common settings. That means they need to test with low-end hardware and sub 5 MBPS internet. But somehow, I doubt they will. A problem with developers now -- be it Google or Microsoft or Apple or web designers -- is myopia: "It works for me, therefore it will work for everyone" or "I like the UI, therefore it is a good design" or something similar.
This post has been deleted by its author
And "no" for exactly the same reasons as to why your Chromebooks never really took off. I just don't want my applications and/or data held to ransom either by yourself, or an internet connection - especially if I am paying to play. If I'm paying £50 for a software title, I'm happy to do an initial install from the internet, but as I only tend to play single player games - I want the software installed locally so if I want to come back to it in a year or so, I know it will still be there.
Also, for well understood and documented reasons I simply don't trust you with any of my data.
So in a nutshell - fuck off Google.
Exactly my thoughts too, they can get stuffed. I very rarely download games from the PSN, I tend to buy hard copies. It's a great feeling when you can look at your huge games stack next to your entertainment systems. It's not the same when you look at a boring list of games online.
Sheesh. I can just picture an album full of titles like:
"I Want It All, And I Want It Yesterday"
"Where's My F^@#ing Unicorn?!"
"The Self-Healing Cake--Eat It And Still Have It"
"If My Call Is So Important To You, Why Aren't You Answering?"
"Help Wanted--Must Be Psychic"
This is bad news guys, and I for sure will not be supporting it. I like going to shops and buying second hand games and picking them up and holding in my hands. This will KILL the second hand games market. Also what happens when the Internet connection disconnects?? I like playing my games offline, it's not always about online play. I have a PS4, when the PSN goes down, everyone playing Destiny and other always online games are stuffed. We will lose control of our gaming. Please guys, don't let this happen, Google are taking to take over EVERYTHING!!
You seriously think there are enough potential subscribers with fast enough internet that the traditional game (disc or download to local hardware) market disappears?
Let's be generous and say 50 percent of gamers have the required internet to use this Google service, the remaining 50 percent aren't going to be ignored by companies wanting to sell them games and hardware.
In any case, some games just aren't suitable for this, such as Street fighter or any VR game.
I've seen fewer comments than I expected "latency makes this impossible" but of course it's a huge issue.
However Google's people are clearly aware of the issue... Lots of the developers are probably gamers for a start. Which makes you sure they have considered it and decided it's not a showstopper. Google may be many things but stupid isn't one of them.
So anyone got any ideas how? Even if rendering and running the game takes no time at all server ping remains. Any they're talking about blockbuster titles not special Stadia games (unreal engine also is supported).
Latency is already an issue in any online multiplayer real time game, and that includes the popular genre First Person Shooters.
The difference is where this latest is manifest. On a local machine, moving the controller moves the cross hairs instantly but the remote player the cross hairs are aimed at might not actually be where your local machine thinks they are.
On a streamed gaming service, there will be latency in moving your crosshairs, but they will be more accurate with regards to the target.
Dave you are of course entirely correct and I had thought about that... what you see on your screen is not authoritative and you see 'skips' when you diverge from what the server decides is correct, if prediction algorithms fail.
But for gamers, surely constantly seeing lag is a big problem - that's WHY developers spend a lot of time making the local client work smoothly and predict the gamestate accurately. I don't know what delay becomes perceptible between me pressing a key and hearing a gun-shot, before it grates?
"pressing a key and hearing a gun-shot, before it grates?"
starts to get annoying past about 100ms on fast twitch style games. (not forgetting the irritating artifact block crap when the video encoder fucks up, and people have already seen that on stadia tests)
driving games are about the same, 100ms+ and corners are hit and miss...(onlive was a joke)
If all you play is candy crush, or slow strategy games you will be all right...this shit ain't for fast twitch games
"Which makes you sure they have considered it and decided it's not a showstopper."
I'm sure they have. That doesn't make their determination correct, though. It will be interesting to see what happens on that score. My prediction: Google hasn't found any technological magic that eliminates this problem. Instead, they'll just carefully choose games that aren't terribly latency-sensitive.
I've stood by as Google shut down service after service, and most recently when they decided to kill inbox - probably the most advanced email client out there for free and paid services alike - I drew the line. I have a few things left to extract, but pretty much have moved away from Google now as an entity for day to day tech needs.
I signed up for the beta of project Stream and got to play Assassins Creed Odyssey from Nov to Jan. I had pretty low expectations going in but was pleasantly surprised that it turned out to be fast and fluid with high quality graphics for almost the entire experience. I did notice some network glitches where the graphics quality went down for a few secs but these were very rare.
Load times were minimal to instant after it got started and I could tell no difference between this and a locally run game. Of course this was a single player type game [a pretty good one] using my old, but hefty, graphic card so the experience could be different with integrated graphics, on a Chromebook, or multiplayer type.
Bottom line was I enjoyed it and would consider it an alternative to Steam games. Another plus was, after the beta wrapped up I received a free copy of the game from Ubisoft to play locally and it plays exactly the same.....
According to tests at Eurogamer, Stadia improved a lot over the Project Stream test run (Assassins Creed Odyssey).
Beside, there is interview with P. Harrison and M. Bakar, people behind whole project. Many technical doubts are explained there, making half of comments here hilarious.
For example, Stadia controller is separate HID unit, directly connected to dedicated client-server and it is not affected of demanding video streaming interaction of data, by sharing same hardware as middle ware. One does not have to be genius to conclude that simple data streaming to/from Stadia controller have priority over incoming video & audio data. This concept minimises lag in interaction.
Regarding those who "prefer physical copy of games"... Well, I adore Tardis style call-boxes, but I am using smartphone in XXI century. It could be that I will play some Stadia games at same, eventually.
A few thoughts:
Latency & streaming resolution limitations (due to limited or inconsistent bandwidth) might turn out not to be a show stopper because for many folk the bar the service needs to clear is like something like "good-enough" only - especially if its priced competitively and super convenient. The history of technology is rife with scenarios where convenient, low cost formats have left higher fidelity ones in the dust (for better or for worse).
Stadia or something similar might well be able to make a compelling selling point out of offering access to higher-than-you-could-otherwise-afford performance. Assuming it works this might turn out to be a no-brainer for consoles but if it scales up to some multiple of the performance of my gaming PC that's interesting too. In some future where this takes off I could see developers targeting a higher spec than anyone can reasonably afford to own (but which google et al can afford to rent to you) and that could change the landscape quite a bit. Enthusiast level PC hardware is already too expensive for large portions of the gaming market – what if the economics of streaming services mean that the baseline ends up being some multiple of today’s high-end such that its basically impossible to replicate that performance level without building your own datacentre?
Depends of course very much on what economies of scale google can achieve with their data centre and also how much performance each user consumes and for how long. If the average user consumes a relatively small amount but pays a flat fee then heavy users (lets say something like casual users playing less demanding games vs hardcore AAA players or something like that) on the same pricing structure will get a very good deal - imagine you bought a super expensive GPU but split the cost with your casual friends who only play peggle occasionally, allowing you to spend hours playing "real" games at a fraction of the cost of owning the hardware on your own. On the other hand if most users are heavy users or if the titles generally require tying up multiple instances to run then you could imagine that the overall cost of using the service will more quickly approach the cost of just buying equivalent local hardware and it might work out to be a much less appealing proposition.
Developers might start developing games with higher latency in mind. This already happened once to some extent with the shift from CRT to modern displays. This might change the sorts of games that get developed (and maybe won't matter for lots of mainstream AAA games where controls have been pretty loose for years) - some might say the traditional rhythm game never really recovered from the death of the CRT & mainstream fighting games seem to be getting more lenient with respect to input accuracy. This is a bit scary but if the trade-off is that game controls get looser in general and in return AI, graphical bells and whistles and resolution and frame-rates all improve I could see it happening. I suppose this will likely preserve a niche for local hardware for less demanding games with tight controls and high skill ceilings – though I wonder if over time this turns out to be a bit of a narrow niche.
As a few folk have already said – many massively popular online multiplayer games already have existing network latency baked into the experience and are doing just fine. It might be the case that even if Google aren’t able to magic the latency away they might manage to move it from one part of the pipeline to the other without making things worse and clear the “good-enough” bar.
Many of us care a lot about their hardware and maintain some sort of notion that we "physically" own our games (mostly now technically not the case) but I don't know that its obvious that the general public at large feel this way so again this might not be much of a show-stopper.
Also updates, configuration, performance tweaking, patching, modding etc - these are a source of "fun" for some number of us but a definite turn off for many folk. If google can actually deliver on their promise of this being a thing where you just click on a link on any device capable of running chrome and more or less immediately start playing that's potentially a strong selling point.
Nice points.
This: imagine you bought a super expensive GPU but split the cost with your casual friends who only play peggle occasionally, allowing you to spend hours playing "real" games at a fraction of the cost of owning the hardware on your own.
By default, this super expensive GPU is not mobile solution. Here, where ever you go, you just have to put Stadia gamepad in your pocket, network speed *somewhere* is only limitation. Which brings other possible problem: Suppose that you want to play some mmo with your gang, but from same spot. Network speed should be multiplied then. Or, same thing described from opposite view: every each of your crew is doomed to play games separated from each other. Actually, that situation would happen with any kind of mmo. Just, Stadia can't be configured easily to replace local LAN-party, as it seems. 5G (mobile) network could be of help here, but... it depends.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
