Competitive games already need to compensate for latency in the network. This system technically won't have any "additional" latency, it will just move where it happens.
The amount of time between you hitting a button and a server somewhere registering the command and sending it to other players should be roughly the same (or close enough not to matter all that much for the vast majority).
Sure, some games where local timing is everything won't work as well (Super Smash Bros probably won't be able to be played on this system at the top tier), but the vast majority of players and games won't ever need that kind of millisecond precision.
It really doesn't take much input lag for things to feel sluggish and unresponsive. I've tried to play Rocket League on a Steam Link, and while it was technically _possible_ to do so, there was just enough sluggishness in the controls that it just felt _off_ - like I was at a disadvantage to every other player. Even something as simple as a 2D platforming game can require precise timing of button presses, and it seems unlikely to me that Stadia or any other similar service will be able to fully address this. I'd love to be proven wrong, however.
I'd imagine a big part is that games which are written for a system like this can do some compensation if they know the latency is between controller and the system (vs games which expect the latency between the client and the server).
I'm also curious if there is anything else going on here than just "streaming video to the user", like allowing some "smearing" to compensate like how VR headsets will do some fancy stuff to compensate for head movement faster than the round trip communication time between the sensors on the headset and the PC.
Latency hiding in "traditional" multiplayer games basically works by doing computations for other players locally (even if it's as simple as dead reckoning and extrapolating a few milliseconds into the 'future'), and only apply corrections when this diverges too much from the server because of network hick-ups. Unless Google invented some new magic, that's not possible with a video stream (but hmm, who knows, maybe they have invented their own 'game-streaming video codec' with additional information for some sort of 2D-dead-reckoning for regions of the video frame... but my guess is they simply don't care about latency).
Yeah, i'm really curious to read more about the tech here.
I know things like the Oculus Rift does some fancy processing to kind of "smear" frames to compensate for head movement while the PC renders a new frame. I'm wondering if this service is going to be doing a bit more than just "streaming video to the user". Like including some metadata and allowing the client to make cheap adjustments to various parts of the scene to compensate for lag?
Either way, I really want to see some people play things on this and talk about how it feels, because I can't imagine Google would just release a gaming platform that's nothing more than streaming video. Especially since it's been tried multiple times before.
Yep, if you did something where local controller input directly impacted the motion vectors for macroblocks in a video frame...you could maybe hide some of the latency, but that seems dubious...
Lets start by getting obvious out of the way - no such thing as competitive gaming on a gamepad. Then you have just encode/decode taking longer than some competitive players whole end-to-end latency.
Why wouldn't you be able to apply the same type of compensation for games like Super Smash Bros? The system (should) know what frame was being displayed when you press your button, and would use that to calculate whether it was a hit or not, rather than the current position.
That's what everyone said about controllers for FPS games, yet there's still console tournaments for Call of Duty and the like. If everyone is playing with the same handicap, it's still a fair match. And sometimes mass appeal is better than playing at the top level.
Some people - strange people - actually prefer a gamepad as an FPS controller, probably because that's what they grew up with. Nobody prefers extra input lag.
This is not a first service that tries to do that. And as before it all ends up with how close the gamer is to the server node, how stable user connection and whether connection is metered or not.
I just pinged my google.com server node - and it's over 16ms. Which means no 60fps.
If that's round trip latency. Then input latency will be 8ms right? Input action to Frame update latencies are reliably 12-13ms+ with a local console routing through a TV.
So if they can do something fancy with rendering frames to the network to sync with your monitor. We might see a minimal increase in actual overall latency.
If all the players are in this service, it means lag is much easier to deal with for developers, since its only input lag. All the players actual compute/rendering would be on a very low latency network.
you no longer have to deal with all the other networking problems between the client software, and server software.
My point is following: you have a character standing in a world. You click "w" - at which point character has to move. Right? On local machine that's what you will get. Not so much with the streaming - first you input goes to the google, being processed there, returns to you and only now you see that character is moving.
But the server knows which frame you pressed the 'w' on. So when it gets the 'w', it rewinds it's state machine X frames (hopefully X==1), applies your input, moves the state machine forward X+1 frames and displays that state.
So there might be a slight visual glitch but you won't be missing your targets.
Yes, even with a Steam Link over Ethernet in my house there's a slight but noticable input lag. Most games are playable but it doesn't feel quite right.
Rocket League is still fun no matter what, even with a little input delay! Turn based games are fine of course. But the only game that has been unplayable for me was Hollow Knight. It requires you to react very quickly to attacks and I just can't get the timing right if I play on the Steam Link.
Surprisingly, a platformer that works decent on Steam Link is N++. It has more fluid controls than Hollow Knight though.
Google is really flexing on people here. With 5G this will be the future of all computing and only google and amazon have the data centers to make it work.
Sundar opened up the event talking about deep learning, I wouldn't be surprised if they used this platform as a way to train their RL models against humans or on data generated by the users.
I'd argue MSFT does, too, and they've already cleared a HUGE hurdle with XBOX (the hardware + game ecosystem) & XBOX Live (social engagement).
The opportunity for Google is huge, but it will require sustained, strong, cross-functional execution that can be challenging for such an engineering driven, compartmentalized organization.
Google really promoting how it's better to develop for than anything else since they can constantly upgrade the hardware. How long before Stadia exclusive content?
They aren't the first or the only one doing this. Crackdown 3 already semi-famously uses Microsoft's cloud systems to do some physics processing in the cloud and not on the client.
Stadia might be a lot easier for devs to code against since everything is in the "cloud", rather than having to split where the processing happens specifically, and so we might see more exclusives there because of that, but MS and friends aren't going to just fall over and give up if this ends up working.
> Crackdown 3 already semi-famously uses Microsoft's cloud systems to do some physics processing in the cloud and not on the client.
Is that true though? I remember in the early Don Mattrick days of Microsoft they were heavily pushing the whole cloud physics thing but ever since then that has been swept under the carpet or heavily downplayed.
I haven't played it myself nor have I done a ton of research, but Crackdown 3 just came out recently, and I heard them talking about how important the cloud aspect of it is to enable the dynamic destruction they have in their multiplayer stuff.
The link below from late 2018 talks about it a bit
So, I work in the games industry and I don't think this will be successful. This has been tried before (multiple times), gamers haven't wanted streaming services. PC gamers already have Steam and this doesn't offer anything compelling over using Steam, and more casual gamers are playing on Ps4 or Switch or mobile. Also, the hardcore PC gamers are never going to accept the latency associated with playing on remote servers.
Google isn't a name that people associate with gaming or gamers, buit they do have a name associated with dropping products after a year or two. I wouldn't trust Google to still be supporting "Stadia" in 5 years time, whereas I do know Steam will still be there.
Also, Google's reputation for "do no evil" has been rapidly eroded in recent times and people are more and more wary about giving this company any more personal data than they already do. Contrast this with Steam, a private company that gamers trust with their data.
I think Google is once again reaching here into a space too late and without anything really compelling to make people switch away from existing Xbox/Windows/Steam ecosystems.
Edit: One other additional aspect to this: it doesn't really matter how fast Google's servers can stream the data, most people's ISPs do not have great consistent connections. Hell, I have Google Fibre at home and occasionally I get issues streaming 1080p Netflix (stuttering etc). Most people have services with bandwidth caps and do not get consistent enough speeds to support streaming games. At my parent's house in the UK, the situation is even worse with lower caps and very often there are disruptions in the connection and so the Netflix stream will turn into a blocky mess for 10 seconds or so.
Yeah, this is my opinion, too, almost right on the nose. Especially the part about only supporting a product for a few years.
I also don't like the idea of games-as-a-service in the sense that I only own a license to the game for as long as I pay a monthly fee. I'm okay with licensing content within the game (i.e. DLC), but I shy away from subscription games services like Xbox Games Pass or EA's Origin subscription.
A streaming game product like this doesn't solve that issue. Additionally, the latency issue, like you pointed out, is too much for me to consider _not_ owning my own gaming hardware.
Have they even said anything about the business model? I'm a few minutes behind on the stream, but I don't recall any mention of licenses or subscriptions. The only "practical" example of launching they game they showed was... out of a button on a YouTube video.
EDIT: A "Stadia store" was briefly mentioned shortly after my comment, it looked like it was tied into Google Play.
The "Stadia store" suggested games would likely be sold or subscribed to. But I definitely have to imagine some sort of trial model being available for the launch-from-YouTube options. Perhaps a time-limited option, that won't stop you from joining a streamer with Crowd Play, but would cut you off after a while if you hadn't already purchased it?
I'm very curious about the expenses game developers will face too, which definitely wasn't brought up. Id Software bragged about just using one Stadia instance, but then later demonstrations showing split screen gaming "without compromise" was shown running each screen on a different Stadia instance. While Google suggests this makes split screen gaming possible again, my guess is that split screen gaming costs the developers twice as much to implement as a service cost, since it uses twice as much hardware. Making it probably mostly a nonstarter anyways.
Gaming PCs need to be upgraded on a 2-4 year basis if you want to keep near the top-of-the-line. One of Google's phones can cost $600-$1000, which is significantly more than a new console. So I'm not sure why you even bring that up, as if there were no economical difference between putting $500+ on your credit card, and trying out a $15-$30 month subscription.
Computer upgrades amortize pretty well. Over the last 8 years I've made 2 upgrades to my computer: A new SSD ($100 for 1TB) and a new GPU which is overkill for many purposes (~600 for a GTX1080). That setup can play many games at higher than 60fps, which beats the hypothetical max of Stadia (and I'll believe "4k at 60fps" when I see it— the Project Stream demo seemed to have a much lower framerate than that while still having noticeable artifacts on a 1440p monitor). Compared to that, $30 a month for streaming (if that's the price) seems as predatory as those companies that let you rent-to-own a TV. Although I'm not a fan of renting games vs owning.
I think we'll have to see what Microsoft is going to do at E3, because they've signalled in the past that they want to take this streaming approach, and they have their own cloud too...
I am a "gamer", I have a 2080Ti and I hope it's the last 1500$ gaming equipment I ever have to buy. It sits idle 95%+ of the time and it is tied to only one of the all the screens in my house.
I knew someone would mention those.. I have two since they launched. However, they don't do 4k or HDR and I still need the 95%-of-the-time idle 1,500$ graphics card.
Steam link games don't have latency compensation since the games are unmodified. I assume that stadia games will have latency compensation, so should be much more playable.
The server will know what frame you were seeing when you pressed your button. So it rewinds it's state machine the appropriate number of frames (hopefully just 1), applies the input, and fast forwards again.
This may result in a visual glitch, but you won't have missed your target.
I wonder if you could use something like the RetroArch rollback solution[1] - render multiple versions of the stream, then pick the closest one based on player input. With a fast enough Internet connection, you could let the _client_ pick, then there is no lag.
Of course, the hardware you have to throw at it to do that quickly becomes ridiculous, unless you can come up with a way of efficiently rendering the multiple versions.
can't speak for Steam Link because I have one, but it was more headache than it was worth to get running.
That said, I had my Bootcamped MBP with a XB1 controller connected over Bluetooth and any input lag was negligible with the Assassin's Creed Odyssey Project Stream demo.
There's no chance whatsoever that Stadia will be lower-latency than Steam's in-home streaming.
No matter how good Google's CDN network is, you can't compete with sub-1ms in-network round trips. There's not really any clever "latency compensation" you can do here.
Even if you rewind state to apply input, which games probably won't support anyway, that still only helps you close the gap added by internet latency. A gap that in-home doesn't even have in the first place.
Were you using old wifi? I've only ever heard of latency issues when using wifi and then only when using older than N. Latency when using Ethernet is less than the HDMI latency on consoles.
You still have an idle graphics card, but you can enjoy the benefits of it from any laptop in your home network. My gaming desktop is wired, my Linux laptop is wifi, and I get 4K @ 60FPS flawlessly.
However, I would prefer not to have to upgrade in the future and offloading the intensive parts to Google would be highly preferable.
Curious: Why'd you go with a 2080Ti and not just a 2080 or 1080? Did you just want the RTX support or did you just want the absolute best possible GPU for gaming regardless of price?
I can't answer for OP, but I went with a 2080ti because it's cheaper than purchasing a 2060/2070/2080 now and buying a 2080ti later. If price wasn't an issue I'd have gone with the best possible Titan RTX.
That doesn't make much sense. If you buy a 2070 now in a few years the successor 2170 (or whatever) will be more powerful than the current 2080ti and cheaper.
Unless you have unlimited budget I find that buying cheaper cards more often will give you better average performance.
The pace has slowed down somewhat. A 2080 (discounting the DLSS and RTX stuff) is about as fast as a 1080Ti, a 2070 is about as fast as a plain 1080, etc...
Why is it tied to only 1 screen? You can use Steam or Nvidia Shield to stream to just about any screen or tablet in the house and it works great. My gaming system is headless and I play all my games streaming to one system or another.
Steam and Nvidia don't have the stellar 11 9s kind of reputation Google has. Also, you have to actually buy those games first, so your library is a tiny fraction of what Stadia offers.
True... but they were talking about how their collection was limited to their single gaming system at home which wasn't true. They have options for that. But both of those are technologies to stream your own games, they do nothing to provide the games themselves. So it really is apples/oranges as a general topic.
Yeah but do you really think you're gonna be like, "You could hand build YouTube at home!" and anyone is going to be like, "Oh okay, WTF Google why even try?"
> Also, you have to actually buy those games first, so your library is a tiny fraction of what Stadia offers.
What makes you think games would just be included in Stadia?
Competing streaming services here are around $10-20/month without any games included. Origin's game access is $15/month without any game playing hardware included.
So... what will stradia end up being, and at what price? Would you pay $30/month for this? When a PS4/Xbox is only $300-400 in the first place, and has a lifespan of ~5 years?
Oh okay, in that case there's no real example you've given for the cost, so $30 may or may not be the cost, but considering a high-end rig approaches $3k and is relevant for 5 years, $30 is a great deal, especially if you get access to a large game library, which is what's happening.
$15/month for a streaming service + $15/month for game access service equals, drum roll, $30/month.
What are you not getting here? Origin is relevant as a reference point for what the games themselves cost on a subscription.
Stadia still needs to actually pay for games, after all. Either as a subscription (reference origin's pricing), or you straight up buy them at the normal $60/pop.
Keep in mind that 2080Ti will give you a way better experience than this will. The actual specs of the hardware here would fall into solidly mid-range. As in, the equivalent entire system would be less than your 2080Ti alone.
So instead of spending $1500 on that piece of gaming equipment spend $250-300 instead. That's what you're actually getting here. A very slow CPU with a mid-range GPU. It's a budget build that's not going to drive 4k@60fps with the bells and whistles. Not remotely close. It'll be fine for 1080p gaming, though, but that's realistically it.
+1, ex networking+tools gamedev here and have almost the exact same thoughts.
They haven't fundamentally solved the latency issue, 50-100ms of ping is brutal and the people who could be a target for this generally won't have a fiber internet connection.
Network gameplay works via misdirection, predictable simulations and knowing how to converge a divergent state(including rewinding in time to re-play hittraces and the like). Streaming gets you none of that.
It'll work for high-latency sensitive singleplayer games but not seeing it scaling out wider.
And when all the clients and the server are sitting in the same DC?
Now you can program your clients and server like they’re all on a LAN. And now you can trust the client. (The client, not the inputs.) Those are huge reductions in complexity. This makes delivering good multiplayer easier, not harder.
But I do agree that input/display latency is going to be a thing. Results may be highly variable and you’ll probably know two people in the same city who have very different experiences with it.
The joystick and display device are just the clients now. You have the same problem just moved around a bit. Without the benefit of being able to use techniques to hide latency, such as client side prediction.
No, the client is still the client. It's still a big bunch of stateful code, but now it's running on the LAN with the server and can't have its bits fiddled by the customer. All that code where the server fastidiously checks the client's untrusted state updates goes away, replaced by a much simpler layer that sanity checks the inputs. All that code where the server has to rewind and replay inputs to deal with late-arriving client state updates goes away (assuming your game isn't cross-DC). All that code where the server is protecting clients from other clients, sending down corrections and fixing up the world so it maintains some semblance of causal order, goes away.
At my day job I'm working on a AAA multiplayer game, and making a properly server-authoritative game with significant client-to-server latency mitigation imposes serious costs in dev time and complexity and security review process. And now, at least in theory, they mostly disappear.
The input/display lag problem is different from what we have now. It's new. We don't, of course, compensate for lag from input to client. (Between client and server, yes, but that's actually a different beast.) And yes, you won't be able to do much about it. You'll be relying on Google to keep your latencies down. They said they're deploying to 7500 nodes, so there's hope. Time will tell. I'm hopeful a lot of people will be reporting single-to-low-double-digit latencies.
But this is much, much more than just moving around the same problem.
I'm used to designing systems and games that work in 100ms+ latent environments since that's what you see in production(or try mobile where it can be 2-5x that).
We were doing 250ms ping times back in '99 with games like SubSpace and the like where you had a predictive game model/design. Yes you lose the overhead of designing for latency because it just won't fundamentally work in those scenarios.
I agree that it will all hinge on what kinds of latencies we see. If Google can produce the same kinds of latencies it does to 8.8.8.8, it's definitely got a shot. Mobile's an even bigger challenge, but I've seen wire-competitive latencies from 5G, so I think there's hope there too.
This does, however, leave VR high and dry. Wonder if they have any ideas about that angle.
Getting a consistent latency to their edge servers is going to be tricky.
Wifi for instance is completely out. All it takes is someone turning on the wrong blender/micowave/power adapter and just the smallest bit of RF pollution will put you into unusable territory. Gets harder as your population density rises.
You can already see this with in-house streaming today. I can stream Netflix but my local LAN XBOne streaming is pretty unusable for things like Monster Hunter.
As your population density lowers the chance of you having the sub-30ms latency w/ 0% loss decreases significantly.
Like I said in the root, they'll able to get this to work for non-action(~200ms window) singleplayer titles, the idea that you could use this for multiplayer is laughable.
I'm seeing anecdotes elsewhere in this thread that wifi worked very well for some testers.
Will it be a worse experience than a 2080 Ti on your desktop? Sure. But will Stadia be good enough to win over the expense and hassle of the alternative? Maybe!
Regarding your last point, I'm not sure Id would stake their reputation on delivering a game that is going to be "laughable" on Stadia. They might know something we don't.
I predict a lot of game teams developing new IP for Stadia will impose a single-DC-match requirement, because the time/complexity tradeoff of implementing all that stuff won't be considered worth it. Then, if their game fails, maybe they'll blame that decision. :-)
I'm not super sure, anymore, about latency being as big a deal. Particularly with regards to monitors. Display lag on a large TV can be upwards of 50ms by itself; display lag on a monitor can be as low as 5ms. If you're viewing this as a competitor to low-end, relatively lightweight console play, the total lag of display plus network is actually maybe not that bad. Also, living in Boston I was seeing 16ms pings on Google's service when I tried it out, which might be an adoption factor. 16ms plus a monitor's display lag is going to be competitive with most HDTVs.
That level of lag definitely makes some games implausible. (Part of why I bought a big-ass OLED is that newer TVs, but particularly OLEDs, seem to have generally consistent display lag; being able to play Tekken 7 on a 55" screen with only 21ms of lag is nice, but to do it on my old LCD is like playing underwater.) But I think it's going to be more practical for more games than the current conventional wisdom suggests.
Yeah, but you hit the nail on the head when you said you have a steady, consistent latency on your TV.
When you're stuffing the pipe w/ 4k encoded content that puts a lot more pressure on the connection and added packet loss or delay is brutal in that case.
All of this stuff is subtle but it compounds in the same way that the average gamer won't be able to immediately spot 30 vs 60 fps but they'll know that the game "feels" better at 60.
I fully agree that it's an inferior good, and I still don't think multiplayer games will be much of a thing. But I do think that even pretty involved action titles (they tested with an Assassin's Creed title) will be okay. Not great, and I'd get mad at them, but I'm also a really demanding user. Most folks kinda aren't.
> they do have a name associated with dropping products after a year or two
Totally a concern. When OnLive closed people lost all their games, no refunds. While if you went to the store and bought a physical disk. It'd still be yours.
The average console quality game is about 50 or 60 bucks nowadays. If you are an active gamer, you could lockup and lose hundreds or even a thousand dollars when they decide to close it.
Most of Google's services are free, but paid ones I'd hope they wouldn't as easily abandon too.
When Project Stream closed down, Google and Ubisoft did give everyone who played a copy of Assassin's Creed Odyssey they could play via Uplay. As much as these companies really don't want to talk about "when we inevitably shutdown", I feel like a lot of player trust would be established if there was a clear commitment to that.
Would you be buying games for $50-$60 through Stadia? What if they follow a model like Playstation Now or Origin Access, where you pay a monthly fee to have access to the entire library (a la Netflix)?
I haven't bought a game on physical disk for a long time. My impression is that in many cases, you're still going to be downloading a big release day patch upon installation. And for multiplayer-heavy games, it doesn't matter if you have a physical disk when the servers are shut off.
Just FYI, but no one outside the tech scene knows about Google's reputation for dropping support of their products or their eroded reputation in regards to do no evil. I have tried talking to people about it and they really just don't give a shit.
Also Microsoft or Sony weren't companies known for gaming when they entered this space and now they are the leaders.
Larry Land even started working on Azure solutions. They were more stable, however, more expensive.
With all that said, there's definitely an interest to be hardware light. So this move into a hosted service should probably be pioneered by the big three hosts: AWS, Azure and Google. They are the biggest consumers of hardware at the moment. And have a hell of a lot more financial backup then Onlive (who only had one product to sell). Xbox has but to partner with their Azure counterparts. Although, I grant you there's probably little incentive due to them being hardware sellers.
All this hardware could have been used for better purposes, of course. Like AI training or Pharma... but like in Stross' Accelerando - it's all going to be used to run economics 2.0 anyway.
> people are more and more wary about giving this company any more personal data
To have the data of every choice and move in every second of every game would be quite useful to calculate the model of the particular user with the level of accuracy barely existed before.
Bungie had to build an infrastructure to push analytics from Halo into their system so they could get "heatmaps" of where players were in a multiplayer map. This fed back into their design process and was used to tweak maps after release.
Building something like that might be easier if all the state is living in a cloud server "right next to" the analytics pipeline, with no need to push state over the external connection. It could, hypothetically, even be built as a library Google supports directly.
That's for sure, but Google wouldn't be Google if they don't invent the way to derivate from that data something like "neural heatmaps" of the particular user to use it for advertising part of business.
Not necessarily. Google doesn't derive advertising (or any other data beyond broad bytecounts, for that matter) from the cloud-customer data stored in their cloud storage infrastructure.
I would definitely try this on some quirky little games that I love to watch the streamers play, but too lazy to download. If the play could start in minutes, I would try it.
This doesn't offer anything more compelling over using Steam? Excuse me?
At this point I see comments like yours in here (from "games industry" insiders) and I see horse breeders decrying the car. "It takes gas instead of grass, it'll never work!"
Problem is Google will never have high quality content from big publisher, it directly compete with their revenue so I'm not sure how they're going to get game on this platform.
They didn't talk about monetization. They could presumably sell Stadia games like Steam games (i.e. there's no subscription). They did mention being able to jump into games quickly, but there are plenty of ways to make that work (e.g. giving everyone some free time).
Ubisoft's revenues are ~$2B. Google's is $130B+. Google already pays Apple $10B+ annually just to be the default search engine on iOS/macOS. You really think there's no room for negotiation here?
Apparently it's a Goldman Sachs estimate. The last time that number was explicitly listed (in court filings) was $1B in 2014. The more apt metric would be what YouTube pays for original content, though I've only seen estimates/projections (a "few hundred million"):
Oh, I don't know. How about the world state saving part? Or what about the ability to instantly jump into the same game as your favorite streamer? Or how about the ability to play a AAA game instantly within 5 seconds without spending hours installing it and then hours patching it. Come to think of it - did you even watch the demo?
The sheer feat of technical complexity in this. Distributed computing, HPC, Networking, hardware, cloud, partnering with other firms etc. It's too complex for me to even imagine the hard work and ingenuity behind it.
Even if the venture fails, I am impressed with the tech part.
I don't think Fuchsia would be used for this.
Vulkan runs on Linux, Unreal/Unity run on Linux, and AMD has a good open source driver. Everything points to Linux being the right choice here.
The only part about it that looks custom is that it's lower spec than what AMD otherwise sells.
Stadia's specs per Google:
Custom AMD GPU with HBM2 memory and 56 compute units capable of 10.7 teraflops
16 GB of RAM with up to 484 GB/s of performance
Radeon MI25's specs:
16GB of HBM2 memory at 484 GB/s with 64 compute units capable of 12.29 TFLOPs
12.29 * (56 / 64) = 10.7.
So this is almost certainly just MI25 chips with 1 or 2 bad 'cores'. Aka, the reject pile. From last generation, at that, since the current gen MI50 & MI60 are both 1TB/s HBM2 memory.
The multi-gpu part maaaaay be custom using the Instinct's extra interconnects instead of what used to be consumer Crossfire. But multi-gpu in consumer space is almost entirely dead, and nothing seems to suggest any developer would do anything for multi-gpu in this platform, either.
That is an interesting detail. I don't want the future of gaming to be Google-based, but it would be awesome if this project - in the couple years before Google ditches it - leads to some real headway in Linux gaming.
Between this and Valve's push for gaming on Linux, another of Linux' Achilles heels is solved. What else is missing before the Linux desktop is adequate for pretty much everyone?
Developing a game that runs on Stadia and one that runs on the customer's local Linux box are probably going to be two different things, unfortunately. :-(
This isn't going to bring any new local games to Linux, rather it "solves" the gaming problem by simply making these streaming games playable on Linux with Chrome, same as on Windows, Mac, on your TV with ChromeCast, tablets, phones etc.
Much like web apps, YouTube, Netflix etc. today already work flawlessly on Linux, so will gaming too with this development. For most people, the browser is already the only "native" desktop application they use on their computers.
Web apps and web video finally made the Linux desktop viable, but also irrelevant in the bigger picture.
It's running on Linux servers and using Linux tech like Vulkan though. Getting a game to run well on Stadia will probably involve a lot of the same steps that are needed to make them run well on Linux natively.
Linux for everyone will happen when I no longer need to explain what a "package manager" or "dependency" is to my retired mother. So, never, basically.
Can I buy just the controller? Sure does seem the PS4 controller is about the best out there for most gaming, and I've been hesitant to buy anything else. This controller seems just as well made.
I've used most of these applications in the past and I'm pretty salty they died. So yeah. I don't have a lot of faith in google keeping services it doesn't find profitable, my guess is the cost of compute will exceed the profit from selling games/game streaming/whatever and then, yes, they will shut it down. Don't act like its not something they don't do or wouldn't do. It's certainly not like making fun of the fat kid at the gym. Or are we turning into hail corporate over here?
One of the biggest reasons to install windows is gaming. If this takes off (or enough games are released on linux as a byproduct) linux could gain a larger install share.
Rewrote this a few times. Overall I think this would be really bad for the gaming world. Not buying consoles would be nice, but the natural business model will be ads or pay per hour or pay per MB. This is going to further drive the industry to towards mass multiplayer and grindy games I think. More accessible but worse content.
Indie shops would probably struggle even more under this model.
The cost per user will vary drastically by the amount they play and how computationally demanding it is. I just don’t see it being feasible without personalized cost. Looking forward to paying minimum cost for minimum graphics too.
It's already failed multiple times. OnLive tried & failed. Nvidia's has been in beta for years. Sony has one.
Everyone in this area has tried this. Nobody has seen what could be described as "success", and the cost models so far have been ludicrous. Turns out renting Xeons and Radeon MI Instincts in a professionally staffed, maintained datacenter is way, way more expensive than a handful of consumer chips in a box in the living room with nobody on-call to monitor it.
The GPU here looks to be basically a slightly cut-down AMD MI25. That'd make a single GPU in this stradia cloud gaming service costs more than 10 xbox one x's. How do you make that price-competitive here?
Yeah, if you know where to look, they left clues about using MI25 hardware. (I haven't been an employee for years, this all unfolded afterwards and, ironically, it is just one search away.)
I'm sure they got bulk/promotional pricing from AMD, plus they're very good at both running hardware with low overhead and packing it efficiently.
> plus they're very good at both running hardware with low overhead and packing it efficiently.
You can't really pack the hardware here since it's latency sensitive. It's straight dedicated resources to an array of VMs. Dedicated CPU cache, even, hence the odd 9.5MB L2+L3 number.
Bulk pricing only gets you so far here. You're still talking gear that's categorically way more expensive than similar performance consumer parts. Not to mention all the other costs in play - data center, power, IT staff, etc...
You can't do time slicing, no, but you can definitely reduce time to first frame in many ways. If you don't do that, you need to provision even more hardware. Packing is also part of the capacity planning phases of a service.
The other costs (power, people, etc.) are amortized over Google's array of services.
Last but not least, it would be very dumb of them not to run batch workloads on these machines when the gaming service is idle. I bet $1000 these puppies run under Borg.
> The other costs (power, people, etc.) are amortized over Google's array of services.
Power doesn't really amortize, and neither does heat.
And capacity still had to increase for this. They didn't just find random GPUs under the table they forgot about, and now that they have a massive fleet of GPUs it's not suddenly going to start handling dremel queries.
This all still costs money. A shitload of it. Someone is going to pay that bill. More ads in YouTube won't really fund gaming sessions. So will this be ad breaks in the game? No way that's cost-effective for the resources used. Straight-subscription model? This seems most likely, but how much and how will you get people to pay for it?
Maybe it wasn't AMD, but they already had a massive fleet of GPUs. It wasn't running Dremel, either. Or maybe they found a way to do that, too, I don't know, but there are already enough workloads at Google to keep GPUs well fed.
I know from experience that Google is very cheap. You tell Urs you saved a million dollars and he'll ask you why you didn't save two. Or five.
If this takes off, the pricing of the service will pay for the hardware (assuming they did a reasonable job there of baking it in). Even if it doesn't, organic growth from other, much larger Google services can make use of the idle hardware.
For the record, I was involved in a couple of projects that required a lot of new hardware. One of them even ended up saving the company a lot of money in a very lucky, definitely unintended way.
>They didn't just find random GPUs under the table they forgot about, and now that they have a massive fleet of GPUs it's not suddenly going to start handling dremel queries.
This strikes me as rather amusing. Google was having such trouble getting their hands on enough GPUs that they decided to build custom hardware accelerators (TPUs) to fill the gaps.
They have TPUs for their AI stuff, and you still have to dedicate these resources while gaming sessions are active. How much monetary value can they really get out of the idle population here to offset the active usage?
You underestimate how much Google tries to squeeze out of all the machines in its fleet. That includes old ones, sometimes to comical effect.
A colleague at my current job told me about utilization targets at Amazon, where he used to work. At Google you could choose to be that wasteful if you really wanted to, but you'd lose headcount. Be more efficient and you'd get more engineers. I.e. you decide if you'd rather get machines or people.
There's also an old paper by Murray Stokely and co. about the fake market that was created to make the most use of all hardware planetwide.
A big difference would be that OnLive had space in 5 colo datacenters in the US. Google has 19 full datacenters around the world and are building more. Plus, google has their own very large fiber network from different POP's and ISP's around the world. The fiber backbone gives them lower, and more predictable latency, compared to multiple upstream ISP's with different connections, issues, etc.
On the other hand, everyone on the east coast (therefore using east coast edge nodes) will be playing from 8-11 EST when the new Wolfenstein game comes out, so how is stuff rationed? Do you make people queue until there is a node close enough to them available? Do you sell the spare GPUs to people in GCP to use for their compute on off times to make up the cash? Do you make it $40 per month?
I think this comment is super underrated. If America is asleep, you can't really use that capacity for players in Europe, since the latency would increase. Likewise if Europe is over capacity, you can't really just assign players to a US server.
And (while I realize you're oversimplifying for the sake of example), it's not just per-continent in this case, but something more akin to per-metro-area.
It'd be a Vega 56 basis not Vega 64 but the problem is that "double the ram" part.
HBM2 memory is super expensive. Like, rumors are 16GB of HBM2 is $320 expensive. Toss in anything custom here and there's zero chance this is under $600/GPU.
Even in the hotly contested consumer market the 16gb HBM2 Radeon VII is $700. And that doesn't have any high speed interconnects to allow for sharing memory with CPU or multi-gpu.
I'm not so sure. To play devil's advocate, this may make games cheaper even without ads. Developers want to not worry about disparate hardware, Google wants you in their ecosystem. Buying google devices, using google services, etc. Content creators want this for better interaction with fans.
I would not be surprised if Google injected some ads into the games as soon as their service reached critical mass and thus killing immersion.
Maybe I'm super negative here but I don't see any pros to this development. Today's games are already dumbed down and steered towards profits only. The only games I play these days are indie games made by 1 person up to a handful of people.
We really don't need more but we need better. (This applies to many other areas too)
I don't understand the arguments being made above.
Pay-per-hour would lead to shorter, more thoughtful games, rather than long grindy ones.
Similarly, monthly subscription where you get access to a list of games actually helps indie games. Being able to jump into an indie game you already own within seconds, compared to having to buy and install a small game you've never heard of before.
I don’t think this is true. It assumes people are economically rational and trying to optimize fun per dollar. But people already should be valuing their time and yet they don’t. If developers must now guarantee players not just try their game but play it for x hours to make a profit, I don’t think they will be pushing for shorter better games.
> Pay-per-hour would lead to shorter, more thoughtful games, rather than long grindy ones.
I'm not sure about that. Currently tons of people pay monthly for grindy MMO games. Expanding that sort of revenue scheme to single player games would encourage devs to create more Skinner boxes to keep players "engaged" over the long run.
> Pay-per-hour would lead to shorter, more thoughtful games, rather than long grindy ones.
Or it ties money directly to length, so you need your game to be 60+ hours to justify it's existence. Or people won't pay that much so we don't get any new dark souls or RDR2 length games.
> Pay-per-hour would lead to shorter, more thoughtful games, rather than long grindy ones.
Pay-per-hour means developers have an incentive to build addictive games that keep you just engaged enough to keep playing for an extended time.
> Similarly, monthly subscription where you get access to a list of games actually helps indie games. Being able to jump into an indie game you already own within seconds, compared to having to buy and install a small game you've never heard of before.
Monthly subscriptions mean that the platform has to choose what games to include and promote based on what is most likely to make users find value in the platform—which means focussing on those with widest appeal, unless their recommender engine can get enough signal to reliably predict niche interest.
Shorter games don't happen. We saw this with Steam when they added a no-questions-asked refund policy for short games. Devs get absolutely punished by making short games now so we're going to see them get longer, and the same will hold if Stadia pays by the hour. If Stadia pays a fixed amount per game we will probably see a proliferation of short games designed to bait people into playing.
> Developers want to not worry about disparate hardware
Most games would be developed for Playstation + XBox + PC + Cloud service to maximise audience (unless one cloud provider gains a monopoly, something so far nobody managed to do in either game platforms or cloud computing).
Of course you can decide to develop only for the cloud platform of your choice, but that's nothing new. Microsoft and Sony already pay you good money to make your game exclusive to their platform (if you're lucky even if you're indy).
I disagree. The natural business model will be a mix between subscription services and free to play games with micro transactions. Microsoft (and others like EA) is already moving in that direction with game pass (monthly subscription gets you all first party games plus partner games).
The biggest concern is going to be internet availability and data caps. I tried the beta for Project Stream (now Stadia) this past winter. It worked well and was impressive, but I have a good internet connection with a 1TB cap. I have friends with much worse speeds and harsher caps and I am not sure if this would be viable for them.
I am more concerned about 'physical' gaming being phased out. I doubt Stadia will do this soon, but I like building a new computer every 4 years for playing games. Maybes it's something I won't actually miss(like how I don't miss CDs for music), but it remains to be seen.
While I am not a fan of many F2P games, their popularity and impact are undeniable.
The trap that you should avoid falling into is assuming that's the only way games can thrive. F2P games are huge and dominate the conversation, but I think indie games are the best they have ever been. My favorite game last year was Into The Breach, and by all accounts it sold well in the same market that Fortnite dominated in.
Games are evolving in weird ways, but it's in a multifacted and diverse way.
FTL was a Kickstarted game from devs with zero pedigree (to my knowledge) and it was also a indie hit. While I agree Into The Breach owes some success to the devs being established now, but I don't think you negate my point about this being a time where games of all types can thrive.
Eh, probably to some degree, but at least anecdotally for me, "new game from FTL dev" got me to watch the trailer, but the concept is 100% what hooked me.
Oh, for sure, I feel exactly the same way. ITB is one of my favorite games in my steam library, and that's the same way I discovered it. The argument I was trying to make is that a lot of people wouldn't have had the opportunity to be hooked had it not been for the "new game from FTL dev"
For some people, the grind is the main attraction, especially if it's well balanced. Look at Path of Exile, that's one hell of a grindy game and people absolutely adore it.
Or worse, a hybrid like Assassin's Creed Odyssey: a $60+ game that gets super grindy, but reminds you that you can opt into micro-payments to get a little experience or coin boost...
Morrowind was way more grindy than Odyssey (make 50 steps, a monster attacks you, repeat until the end of game). Odyssey just made all enemies within +/-2 levels of your own character, contrary to Origins, where levels were preset for each area.
Game Pass works because you're essentially temporarily unlocking a library of games to be downloaded to your device. It really isn't an indicator they're moving in the same direction as Stadia.
It is kind of another beast altogether for Google, Shadow, Microsoft (xCloud), etc. to dedicate actual hardware for your usage. Shadow and others like it are essentially remote VM w/ GPU that you rent by the hour. Stadia, xCloud, etc. we don't know what the business model is going to look like. The only positive here is that Microsoft, Google, and Amazon are all cloud infrastructure companies so in theory, they could price their offerings cheaper than a company that is a tenant on their systems.
It's not just game pass though. Microsoft has flat out said they are moving towards streaming. They are also doing things like making Xbox live a service on the Nintendo Switch, iOS and Android. Game Pass the way it is today is just an indicator into how things will be.
Microsoft has made clear that they want Xbox to be a platform independent of the physical box they sell to people. They are 100% going in the direction of Stadia. The big question is how they are going to do it in regards to their next console.
They have announced xCloud, yes, but have they announced the business model/pricing? So far, pretty sure the answer is no. So it may be a subscription, but it may also end up being akin to renting a VM with GPU in Azure. Time will tell.
> The only positive here is that Microsoft, Google, and Amazon are all cloud infrastructure companies so in theory, they could price their offerings cheaper than a company that is a tenant on their systems.
That would be considered unfair competition and they would probably be fined by EU, I guess
I agree with your comments. I'd also add Stadia could operate a traditional platform/store model as well. It'd be a big ask for big publishers to launch their huge-budget flagship content straight into a content bundle. But you could have them sell access to people with Stadia accounts same as they sell physical/digital copies to Xbox/PS/PC owners.
Imagine simply moving all hardware support issues "in-house" or at least trading in all of the pimply teenagers with hardware issues for a single team of engineers working for one of the smartest companies in the world.
The vast majority of your support department can be let go.
I'm struggling to understand how fewer casual gamers buying consoles would impact indie developers. Could you expand on your point?
Indie developers don't get a cut of console sales, but they do care about how many people buy/play their games, and if this widens the market it seems like it'd be a win for them to me.
Most indie games are short. If money is allocated to content producers based on play time, I would bet that most would earn less than a typical indie game purchase price per user.
That’s just speculation mind you. Things like Stardew Valley would probably do even better.
Nowhere did it say it's per MB or hour or whatever. Yet it's the top comment here. It's just fear mongering and hearsay on your part when most likely it will be a fixed monthly price.
The cost will change per game, so it could be that game's cost is associated with how much compute they need. I think it will probably be a monthly subscription based on the gym membership model. Most people won't use it a lot and they'll make all their money on them.
There's the major difference that you can't "pirate" a single-player game that is only available remotely, and that you can access such games remotely from any hardware.
So it could result in a renaissance of the classic AAA 3D single-player games, perhaps.
Writing "ps4 piracy" and "xbox one piracy" into google is a whole lot shorter than your comment. Then again, it doesn't give karma like a cool retort would.
I think undoubtedly, I would bet on the longevity of a streaming service from Valve or Amazon more than I would one from Google. But Google's past history of killing free side services doesn't offer the most relevant precedence. This is a service into a mainstream (and growing) industry that gels with Google's strengths, particularly scalability and AI. YouTube Premium/TV would be better precedent, though the jury is still out on those. I think there's still a difference, though, in that YT+/TV entered a market with already dominant leaders (traditional cable, Netflix, free YouTube). If Stadia can satisfactorily mitigate the problems inherent to streaming, and get it out before Microsoft's rumored Xbox development, it will be in a prominent position in a new field for gaming.
This is going to be an alternative for some types of games.
And quite on the contrary I think it would awesome for indies, since they tends to be pretty short. And it is much easier for them to gain consumers, if they went viral over streaming, many of which are, like horror games.
Why are you sky-is-falling this announcement? Where in their release did they say anything about per-MB or ad based business models? Did they say anything at all about personalized costs in general?
No, they didn't, and every single thing you've suggested about their gaming platform is also true about every other streaming platform, and yet none, literally none, of those platforms have tried any of the things you're complaining about.
You basically just made up a bunch of things to complain about because nothing in the actual release was objectionable. Yours is not a helpful way to react, my friend, though it is a popular one.
I mean, I’ll give you that nothing is confirmed, but nobody has made this work yet. It’s mostly tech testing, so current business models aren’t a good metric.
I’m open to debate about how this could be priced, but I’m pretty comfortable pointing to existing cloud computing business models or streaming services as a precedent.
I would invite you to come up with an alternative business model for serving people who like to play high end graphics games for many hours a day.
> but I’m pretty comfortable pointing to existing cloud computing business models or streaming services as a precedent.
Why are you this comfortable? Netflix, YouTube TV, Twitch, Sling, Amazon Prime Video -- basically all streaming services offer flat rates, not per-MB ones.
Further, all existing game library services tout unlimited gaming as a primary selling point! That's the primary reason you opt into Gamefly or Nvidia Shield, at least according to their own marketing.
And this offering is not for high-end gamers. It's taking the benefits high-end gamers get for their investment into their hardware, and making it available to the millions of more casual gamers. This isn't for high-end gamers, so creating a business model for them using Stadia makes no sense.
Finally, you're not thinking of this at the right layer if you're thinking in terms of things like s3, ec2, lambda, etc.. This is the product that's built on top of those, and the single price problem has been present for hundreds of years. It's a solved one, just ask any current MMO or hell, any clothing manufacturer. You're basically saying that an XL t-shirt is going to cost the same as a S t-shirt, despite tens of thousands of examples to the contrary.
> Netflix, YouTube TV, Twitch, Sling, Amazon Prime Video -- basically all streaming services offer flat rates, not per-MB ones.
YouTube (and Google Play TV & Movies, which appears to carry the same for-sale/rent content in a different storefront) and Amazon Video also both offer purchase of individual content items as well as a common flat rate subscription to certain content.
The amount of variable cost to stream music or video to a user is significantly less than the the variable cost to render high end graphics. The high end hardware costs money and depreciates. Why would it cost less to rent a gpu hour for generic purposes than it would to rent a gpu hour to play video games?
... no, not everything is scalable, nor does everything benefit from economies of scale.
This does benefit from economies of scale, but it’s not something you can just solve with infrastructure and fixed cost. No matter how many computers you have, you’re still going to do multiple orders of magnitude more calculations to render high end games than stream a song. And you’re going to deal with difficult load balances because every twelve year old gets home from school at the same time (exaggeration but point stands). GPU time costs money.
This benefits massively from economies of scale, and yes it is very much something you can solve with infrastructure and fixed cost. Fundamentally it doesn't matter if you're streaming a song or streaming a game.
GPU time costs money but it's a fixed cost, doesn't matter what the GPU time is being used for, therefore it won't be per-game. The end.
It’s only a fixed cost if the computers must be running at all times. That’s not the case. GPUs consume power, and require energy to cool. They also burn out over time and need to be replaced.
For similar reasons, you’re not just going to rake in the money mining bitcoin because you bought a bunch of computers.
Or maybe to make the point even stupider, you could make a game about training neural networks same as you would on a real cloud service provider. If you can understand why google doesn’t charge a simple monthly flat fee for cloud computing of neural nets, you can understand why they can’t charge a simple monthly fee for computing neural nets in a game.
The whole gaming industry is positioning itself for cloud-only streaming in the next 10 years. With 5G+ everywhere, reduced latencies, nearby datacenters full of specialized GPUs, many gaming studios plan to release streaming games only at some point, cutting off local players completely (it makes a perfect business sense, collecting regular monthly rent for gaming). Unless there is some regulatory pressure, the only way to play AAA games in the future will be via streaming, local computers will be pretty dumb latency-optimized streaming machines.
Not sure how would that affect game developers though as there won't be any attractive way to programming for young people; before, making an own game was quite an attraction to jump into programming and a motivation to study hard.
I think my concern about this wholly depends on if it completely takes over the business model. I really want to own my games. But, I don't mind if alternate streaming models exist, just if they become seriously required.
It’s Google, which is notorious for smothering virtually everything unrelated to search or ads in the crib, and especially when it’s something they made in-house. Add to that streaming games; problematic for the (by far) most popular two genres: huge multiplayer shooters, and sports. I see almost no chance of this succeeding long-term, and I’d put the odds of it going EOL within 5 years at over 80%.
These huge companies will keep throwing money at streaming because they see gold at the end of the rainbow, but I’ve seen no indication that they have a plan without giant question marks before the “profit” step. Look at YouTube, something Google acquired in 2006, is dominant in its space, and still doesn’t make them money. Yet somehow games, which are more finicky and reliant on universally good internet connections is going to work for them?
This is trend-chasing, once again without any real new ideas to overcome existing challenges.
> It’s Google, which is notorious for smothering virtually everything unrelated to search or ads in the crib.
It's apparently deeply tied to YouTube, and particularly to help provide an onramp for more (often ad-opportunity-generating) gaming content on YouTube a service that is both an ad-platform and the venue for at least two distinct revenue generating premium services (the older of which is ad free) that Google has not strangled in the crib.
There's so much that this will hurt in the gaming industry. Some of my thoughts:
1. As subscription services take over, the upfront revenue game studios see will drop. This is just simple math: Xbox Game Pass costs $10/month, which means its the total cost of a AAA game over 6 months. In a traditional model, many gamers would expect to buy, lets say, 2 AAA games per year. In this new model, I can play as many as I want. And even if I only play 1 or 2 every year, I'm almost definitely going to be "dabbling" in the collection for other games I may want to play, ESPECIALLY if they're instant-on like Stadia. Even if they pay-out to studios based on some metric derived from time spent in game, there's no way studios will get the same level of income as they did before. (note: this is exactly why Spotify is having such a hard time, and why they're branching out beyond music. royalties abstracted behind a subscription service suck for the bottom line)
2. So upfront revenue drops. How do studios make that up? In-game transactions. They're already huge, and they'll just keep getting bigger.
3. So what, micro transactions (mtx) are the "new normal". Well, the top 5% of games can afford that decreased upfront revenue by making it up in mtx (think: Fortnite, Apex Legends, CoD). The trailing 95% can't (think: Indie titles).
4. Beyond that, you can bet your bottom dollar that Stadia will pay out tons to the AAA studios just to get their names on the platform, given that Google has no first party studios to speak of. Assassins Creed gets enough upfront revenue to make it worth their while, meanwhile the next indie darling is left out to dry, further balkanizing the gaming industry
5. Switching gears: A massive number of software engineers in the industry entered it because of gaming. Games, even back in the 90s, were such a clear application and value of computers that it was obvious, even to children, that they'd be something huge. It inspired a generation, to not only play, but to mod and even make their own. Now, we're moving that all off into the cloud, hidden from the next generation. Google wants you to own a Chromebook and consume their products, not understand how they work.
6. Speaking of modding: Its literally the source of the world's most popular games. Battle Royale? You can trace its roots back to mods for ARMA and Minecraft. MOBA? Dota, a mod for warcraft. Creativity happens in environments that large corporations can't recreate, and traditionally a great platform has been starting with a base game, a great game, that some studio created, then exerting your creativity on top of that platform. It benefited everyone, including large AAA studios who could then copy your idea and make millions. Yeah, good luck modding on a blackboxed server a hundred miles away.
7. But fine. I guess we're moving into the future and this is part of it. Except, there are millions, even BILLIONS, of people around the world without the internet capability to even join this service. Google tried to help solve this with Fiber, and gave up. Its fucking hard. They'd rather do easy, cool things, like cloud gaming. Modern consoles are bad enough; my brother, who lives just an hour outside of a top 10 US city, recently told me that he downloaded Fortnite on Xbox for the kids. It took a week of 24/7 downloading. Most games drop with day 1 patches in the dozens of gigabytes, even if you buy the disk in-stores. The sheer arrogance of Google, to get up on stage and claim this service is gaming for EVERYONE, the apex of accessibility, is disgusting to me. They're stuck so far up their own ass they've become the ouroboros.
8. Well, streaming games have taken over the world. Let's say you want to compete with Google on this streaming game front. All of the top three cloud providers now want to get into game streaming. So, no way can you compete on cost there with them; they own the data centers and give their game streaming div...
You hit every single point I tried to make to my friend about half an hour ago. On one hand, I'm terrified that all of this will come to pass; on the other, if this is like any of Google's other projects, it's going to evaporate in two years anyways. Whatever happened to OnLive?
It wasn't the right time for OnLive. The tech wasn't there, in networking infrastructure (the internet), compute infrastructure (DC tech), or hardware (graphics cards).
To that last point; it is impossible to understate how fundamentally important Nvidia's Pascal architecture has been to the development of both gaming and AI. In my mind, its the most important computing chipset invented in the 2010s, and belongs among the "world's greatest" chips next to the Intel Core architecture, Apple's A-series, Pentium, and the 8086. It put Nvidia, quite literally, 5 years ahead of the competition almost overnight; AMD is still catching up, three years later, to the perf-per-dollar and perf-per-watt of the GTX 1080.
That chipset, and the cards that were made with it (GTX 1080/1080Ti namely) were the first indication that DC rendering with a stream-to-client architecture was actually possible for video gaming. Before that it was hard to make an economic case for it.
OnLive was purchased by Sony, and its easy to conclude that they repurposed their tech for their Playstation Now service. So it lives on.
This is such a parochial and short-sighted view. Let's block new technology because "game studios will die". Businesses need to adapt. And if people still like games that don't rely on microtransactions or multiplayer, studios will make them. There is a place for both.
Also, to point 7. guess what billions of people can't afford a console or an expensive PC rig. Yet, in developing countries data is already very cheap if not free. So, this DEFINITELY is a big step closer to unlocking games for them.
Do you really think data is delivered to people in those countries in any sort of latency that would make a service like this remotely enjoyable to use? A massive number of people in the US don't even have internet that could support something like Stadia, let alone developing countries.
By comparison, shipping "edge devices" (aka, uh, COMPUTERS) running an 845 Snapdragon or Tegra (like the Switch) is cheap and getting cheaper. What makes more sense: asking someone in a developing country to pay $200 one time for a general purpose computer useful for everything including 1080p gaming, or $10/month in addition to, uh guess what, some computing device they'd already have to own to access Stadia.
The adage goes: never underestimate the bandwidth of a station wagon full of tapes barrelling down the highway. I mean, the follow up is usually "but never forget the latency", though in this case that doesn't apply. Point being: the internet isn't the answer to everything, but if the only tool Google has ever known is a hammer then every problem is going to look like a nail.
honestly none of this even matters until they can offer a service with comparable input latency to local play, and so far every indication is that they can't. the whole service falls apart if it feels like shit to play.
Unless they've solve the speed of light issue then we don't need to wait for reviews. Even using a remote desktop on the other side of the city isn't a pleasant thing to be doing full time and that's not remotely as twitch based as gaming.
No they didn't, that was simply a bandwidth issue, we've been streaming video since TV was invented and with the right equipment you could stream video over the internet before the web even existed. No equipment exists that can alter the fundamental speed limit of the universe.
Yep. This is the kicker. For streaming videos, only bandwidth matters but for video games, both bandwidth and latency matter.
Reminds me a bit of the 30/60 fps fights a few years ago. Sure, 30fps games look more "cinematic" and 60fps movies look uncanny, but 30fps games feel less responsive.
Google Cloud has regional US DCs in western Iowa and central South Carolina. A midpoint between those two locations roughly lands on Nashville TN, which is ~600 miles away from either. Light could make a roundtrip of that distance in 6ms. Of course, the internet doesn't allow for latency at the speed of light, but that's the physical limit, and that's plenty; a typical internet browser alone has input lag of 10ms [1]. In order to achieve 60fps, frames have 16ms to be rendered.
But the regional DC is only the worst-case, because they've said they're deploying these things in 7500 locations around the world. That's unprecedented scale for a tier 1 cloud provider at the edge. They know that they have to be close to consumer populations.
Also consider this: Once cloud streaming takes off, we're going to see deeper integration into the frameworks and game engines themselves. Imagine a game engine which is built for streaming. It could do input prediction, doing a "light rendering pass" of frames for N possible inputs the input buffer might receive on the next frame, before it receives them. These custom chips they use have plenty of headroom to do this at 1080p, and most controllers have, what, 12 buttons + all of the joystick states? Depending on the game this might be possible (example, hard to do in multiplayer). Combine that with the natural advantage a cloud-hosted multiplayer game would have in networking with other clients to resolve game-state, and you can see that its not just a strict downgrade; it might be possible that we'll see improvements in the performance of games beyond just the typical "new year better graphics" cycle.
I watch WebRTC streams from California in Germany with 100ms network latency. Speed of light has never been an issue. Latency within the same continent or within the same city is much lower.
Why on earth would it be pay per hour or per MB? Does Netflix charge that way? How about Spotify?
Time has shown that the best revenue model for these kinds of services is finding a monthly price, or multiple tiers of a monthly price, that adequately cover the costs incurred from heavy users and light users alike.
For end users, I agree. It’s also interesting to consider how content providers will be paid: per user who plays, per user-hour, a flat catalog fee, etc., and what downstream effects this will have on games.
Netflix and Spotify’s cost structure are different. Streaming content is cheap. Rendering a high end video game can get pricey because you need GPU time. There’s a lot of other considerations too. Like, a lot of people will pause a stream and let it sit for a few hours. That’s fine. What if people want to leave a game running for a few hours? The computational burden is still high.
Why would they charge per hour or per megabyte? That reduces stickiness. Much more likely is that they'll just get a cut of the game developers' sales, just like the game console vendors do. They'll set the cut so that the net profit pencils out over the long term.
I give this 2 years before Google removes all references of "Stadia" from it's product lineup. They have never stuck with a product which, relative to rest of their products, has smaller but dedicated adoption.
For example, Valve will keep updating/supporting/hosting events of CSGO even if it has <1Mil active users on PC, but I can totally see Google pulling plug on it and burn dedicated followers.
> For example, Valve will keep updating/supporting/hosting events of CSGO even if it has <1Mil active users on PC, but I can totally see Google pulling plug on it and burn dedicated followers.
Bad example because CSGO has been sitting at #2/#3 on steam's active player charts for years and years. Consistently ~10x the player count of, say, Rainbow Six Siege. Continuing to invest resources into a game with CSGO's player count is just all around solid business strategy, particularly since it has ~500k concurrent users at any given point. Active player count would be way, way over 1Mil.
TF2 would be a better example, but I don't know how much (if anything?) Valve is still doing with the small TF2 community. It still has a respectable 50k concurrent user population, but has Valve been ignoring them? Are there still updates for it?
Particularly if Facebook overtops them with some sort of streaming Oculus appliance. This is already a killer app for people on the edge about upgrading older PCs. I beta tested this as did a friend - he was able to play on a chromebook! But if I could play Elite: Dangerous on a streaming VR console I'm not sure I'd ever buy another high end PC again.
Speed is not the (main) problem. It's latency and jitter. With the last one especially bad on mobile networks. To get at most 1ms of additional delay you also need a datacenter every 300 km.
I used the precursor of Stadia, Project Stream. There were moments where the video went fuzzy, but for >95% of the time, it ran fantastic. My gaming PC setup is an i5/16GB RAM/GTX 1080 and it stutters more than the stream did (I assume that's on the CPU side). I'm sure Google has the compression algorithm chops to make this all work.
Unfortunately for me at least, what I want is more akin to a cloud VM where they can host my games and I stream them wherever, which this isn't it. I've already got games between Steam, uPlay, Origin, etc. I have zero interest in buying a game 2x just to be able to have more portable play.
I'm pretty sure you won't buy them (again). You will rent them. Anything but subscription-based pricing model for Google Stadia would be a surprise to me.
Google operates tiny little datacenters in every ISP POP in America. That is why the video I just watched on Youtube in Oakland came from some place 2ms away even though Google's nearest real datacenter is in Oregon.
Video files (even live stream HTTP based protocols like HLS) can be cached very easily on CDN infrastructure.
However, a gaming session means to have a dedicated daemon running for you somewhere. I doubt this can be deployed anywhere on the spot, but perhaps they have some amazing technology for that.
How do you edge cache games though? Edge caching youtube requires a giant pile of hard drives, Edge caching games requires a giant pile of hard drives and a giant pile of compute power.
It's not the edge cache itself that matters; it's that many ISPs peer directly with Google.
Google Cloud can route the audio/video/keyboard packets mostly over Google's private network and then only use the public internet once it gets to your ISP (or their transit provider). This provides Google with more control over how the packet gets to the end user.
Google provides a similar service to Google Cloud customers as the "Premium Network Service Tier".
Huh, that's actually really interesting. I was going to say, "but even then you're still dealing with the speed of light", but I guess at the limit, it takes 4ms to reach 1,000 miles, which should be enough to get most anywhere in the U.S. to a Google data center. The latency dynamics here are a lot less implausible than I was suspecting at first, given peering at the local ISP level.
Gaikai was one of OnLive's competitors. It had better management and probably better tech as well. Sony bought it out and it still lives as Playstation Now. I believe Dave Perry is one of the founders.
OnLive was both before the Cloud era (where they manually had to maintain many points of presence that were near consumers and they didn't have the money to do it near what Google, Amazon and Microsoft are able to do) and before GPUs were really mainstream for in normal data center servers to the scale this type of service requires.
> the company had deployed thousands of servers that were sitting unused, and only ever had 1,600 concurrent users of the service worldwide
They were burning through all of their money because they highly overestimated the audience. That's one of the main problems the cloud was made to solve.
This particular cloud will only solve the problem if google will find a way to sell the time of these AMD GPUs to someone else.
Traditional GPGPU customers, and machine learners, usually use CUDA instead of OpenCL / VulkanCompute / DirectCompute. At least from my position is looks that way. I’m a freelance developer and my clients are picking CUDA in ~75% cases. The rest is DirectCompute, if computing on client PCs and hardware compatibility is required.
Right partially because OnLive couldn't spend the money required to put servers close enough to give users a good experience. It would have taken a monumental amount of money for them to expand in a way that would facilitate that, and custom make servers that had consumer grade GPUs in them.
Furthermore, there are other technological advances that have come even in the past 2 years that help facilitate this, like Secure Reliable Transport protocol (https://www.srtalliance.org/).
Things were just too stacked against OnLive at the time to make it work, but most of those barriers have since been removed after they had gone under.
So, who wants to make a bet on how long this "experiment" will last? I'm thinking 2 years before Google gets tired of it and abandons it, and another year before it's taken behind the shed.
It's not ironic at all. VR movies were always experimental and it looks like they realized there isn't a substantial market there, video games on the other hand are a proven industry that they've just entered.
The difference is they launch high-profile, promising-the-moon projects and put their full weight behind them, and then give up a couple short years later for inscrutable reasons ([cough] Google Fiber [cough])
They lack the ability to commit to a space and cultivate it, casually changing course on a whim and leaving masses of people spinning in their wake. Just look at how they've managed YouTube and its guidelines, bans, etc. Or the Play Store and all of its debacles. Were I a game developer, I would never want to be beholden to a Google gaming platform.
> They lack the ability to commit to a space and cultivate it
That's because states, entrenched providers and in some cases local governments made every inch of work cost a mile of time. The cost was making Fiber a non-starter... ISP's benefited from nationwide networks largely funded by governments, then used their monopoly influence to prevent any competition, and "hired" state governments to enforce their monopolies.
But we can hand-wave all of that away and claim it was Google's lack of commitment.
I don't deny that ISP's make it far harder than it needs to be. They make it impossible for small upstarts. But Google, of all companies, had plenty of resources to make it happen if they really wanted to.
Blowing millions of dollars on an investment that will take decades to pay off isn't good business. Google decided, rightfully, that the answer was webpass. It's a better revenue generator, and has considerably lower barrier to entry.
The point is they overcommit and then back out instead of doing more careful due-diligence. Google Glass is an example of the right way to explore a new product; most of their ventures aren't so conservative.
Google Fiber wasn't really that inscrutable. It's clearly outside of their core competency, it required collaboration and agreement with local governments, as well as construction projects, and to AT&T's snarkily made point, it's way harder than a typical software project.
This is closer to their existing product portfolio, it doesn't have nearly as a difficult upstart cost relative to Fiber, and the business models are pretty clear. I think a closer analog would be the rumored reductions in their hardware division, but given that they're still making successful hardware + software, it could be more about reorganizing around priorities than it is an indication of their reduced investment
There likely would be no offline option for this if (when? Ha!) it gets shut down.
It’s not like a photo site where you can just download your content. Whatever is built to run on Stadia will not run on your home machine or anywhere else.
The failure rate of startups is drastically higher than Google products, so it's only unusual in the sense that they have way more success than the industry as a whole.
It's less about them killing products and more about them following the same frustrating sequence of steps (for products that people actually liked):
1. Release a product, make it decently good.
2. Kill it in a few years after it has gotten widely adopted.
3. Release an alternative product that does the same thing, but, a lot of times, worse than the original and/or with missing functionality from the old one that people actually care about.
4. Bonus step: make it really confusing (looking at you, Hangouts => Allo/Duo).
5. Repeat the cycle.
Notable examples of it that come to mind:
* Messaging: GChat => Hangouts => Allo/Duo => I got tired of keeping up with that they are pushing now
* Music streaming: Google Music => Google Play Music [just a rename, I think] => YouTube Remix => YouTube Music
* First party Android experience: Google Play Edition => [I feel like I am missing something in-between here] => Android One + Android Go
See, that's part of the problem. As an end-user, to me all those sound kinda samey and fill the same niche, but technically are different products.
As for others, I personally stopped using non-essential google products (i.e., the ones I believe that Google is unlikely to screw up without completely destroying faith in the company, like GMail or Maps), so I can't come up with anything. If you are curious, you can check out https://killedbygoogle.com/ and see if there is anything there that might fit the bill.
Disclaimer: I have zero affiliation with that website. I found it by trying to look up a list of killed off Google products, hoping it would bring up some memories. Then I realized that there are way too many products on the list for me to parse through, and the website has a nice and clean UI/UX that I am not ashamed to share here.
They might roll it up into youtube or something, but the main components are cloud compute (GCP) and video streaming (Youtube). Its naturally aligned with their other initiatives so its pretty safe I would think.
It was only a matter of time before this tired and predictable comment appeared. I'm surprised he didn't mention his withdrawal symptoms from the shutdown of Google Reader which he probably never used.
As for your bet, I'll take it. Let's say $100 USD?
I'd say one promotion cycle for the top executives on the product, 2-3 more for the next tier of engineers/product folks to ship some cool stuff, then a year or two for the product to coast before no one wants to take on the technical debt. So I'll predict its shutdown will be announced by July 2022.
From a preservationist angle, this would be a terrible path for games to go down. Any exclusive games released using Stadia would be impossible to play once the servers shut down because users don't have any access to the software.
Even if MMOs shut down, there is usually an opportunity to write a new server by reverse engineering the client. I can't really support anything about game streaming unless it was tied into an actual game you could install or play without relying on their servers to exist.
This is what I've seen everything trending towards for almost ten years now. The next step is to build hardware that utilize the full benefits of data center gaming. Vr headsets can shrink and are only limited by the method of getting light into your retina. Same goes with any kind of haptic feedback or audio feedback.
I'm skeptical they'll be able to overcome the latency problem. At some point there's a hard cap that only infrastructure can help with - not software - and Google's already given up on rolling out their own infrastructure.
Given the number products and services that Google has shuttered over the years and their focus on learning all there is about their users, I am going to sit this one out.
There is definitely some cool work here though: being able to have access to a library of games without lugging around a console or computer is an interesting idea with both pros and cons. The pro that I can think of is not having to be responsible for upgrades. The con is that we continue to move towards a world where we rent more and own less.
Probably worth noting: this project appears aligned with Google's Cloud and mobile initiatives. It's unproven, but it's originating from spaces where Google's products have stuck around.
576 comments
[ 2.8 ms ] story [ 281 ms ] threadReally curious what the network performance requirements would be for 4K or 8K game streaming. Haven't heard them admit to that.
Competitive games already need to compensate for latency in the network. This system technically won't have any "additional" latency, it will just move where it happens.
The amount of time between you hitting a button and a server somewhere registering the command and sending it to other players should be roughly the same (or close enough not to matter all that much for the vast majority).
Sure, some games where local timing is everything won't work as well (Super Smash Bros probably won't be able to be played on this system at the top tier), but the vast majority of players and games won't ever need that kind of millisecond precision.
I'd imagine a big part is that games which are written for a system like this can do some compensation if they know the latency is between controller and the system (vs games which expect the latency between the client and the server).
I'm also curious if there is anything else going on here than just "streaming video to the user", like allowing some "smearing" to compensate like how VR headsets will do some fancy stuff to compensate for head movement faster than the round trip communication time between the sensors on the headset and the PC.
I know things like the Oculus Rift does some fancy processing to kind of "smear" frames to compensate for head movement while the PC renders a new frame. I'm wondering if this service is going to be doing a bit more than just "streaming video to the user". Like including some metadata and allowing the client to make cheap adjustments to various parts of the scene to compensate for lag?
Either way, I really want to see some people play things on this and talk about how it feels, because I can't imagine Google would just release a gaming platform that's nothing more than streaming video. Especially since it's been tried multiple times before.
That is, in fact, how video compressors work. It's called motion estimation.
Some games like Rocket League and many fighting games are primarily played on a controller or some kind of gamepad at the top tier.
I just pinged my google.com server node - and it's over 16ms. Which means no 60fps.
Besides - low input latency means you get response that is not exactly what you are doing in the moment.
So if they can do something fancy with rendering frames to the network to sync with your monitor. We might see a minimal increase in actual overall latency.
you no longer have to deal with all the other networking problems between the client software, and server software.
So there might be a slight visual glitch but you won't be missing your targets.
There will be a visual glitch every time you press a key.
Actually I don't even think this will work properly at all. You can do some prediction, but latency still kills you.
The moment we're talking about CSGO or any decently fast paced shooter and it's a no-go for sure.
Surprisingly, a platformer that works decent on Steam Link is N++. It has more fluid controls than Hollow Knight though.
Sundar opened up the event talking about deep learning, I wouldn't be surprised if they used this platform as a way to train their RL models against humans or on data generated by the users.
The youtube integration is also a shot at twitch.
The opportunity for Google is huge, but it will require sustained, strong, cross-functional execution that can be challenging for such an engineering driven, compartmentalized organization.
With Android, Chrome, Chromecast, Youtube and GCP they have all of the pieces in place.
Stadia might be a lot easier for devs to code against since everything is in the "cloud", rather than having to split where the processing happens specifically, and so we might see more exclusives there because of that, but MS and friends aren't going to just fall over and give up if this ends up working.
Is that true though? I remember in the early Don Mattrick days of Microsoft they were heavily pushing the whole cloud physics thing but ever since then that has been swept under the carpet or heavily downplayed.
The link below from late 2018 talks about it a bit
https://www.windowscentral.com/inside-crackdown-3s-azure-clo...
Digital Foundry's opinion:
> Crackdown 3 doesn't deliver much of the promise previously showcased from the power of the cloud.
https://www.eurogamer.net/articles/digitalfoundry-2019-crack...
Google isn't a name that people associate with gaming or gamers, buit they do have a name associated with dropping products after a year or two. I wouldn't trust Google to still be supporting "Stadia" in 5 years time, whereas I do know Steam will still be there.
Also, Google's reputation for "do no evil" has been rapidly eroded in recent times and people are more and more wary about giving this company any more personal data than they already do. Contrast this with Steam, a private company that gamers trust with their data.
I think Google is once again reaching here into a space too late and without anything really compelling to make people switch away from existing Xbox/Windows/Steam ecosystems.
Edit: One other additional aspect to this: it doesn't really matter how fast Google's servers can stream the data, most people's ISPs do not have great consistent connections. Hell, I have Google Fibre at home and occasionally I get issues streaming 1080p Netflix (stuttering etc). Most people have services with bandwidth caps and do not get consistent enough speeds to support streaming games. At my parent's house in the UK, the situation is even worse with lower caps and very often there are disruptions in the connection and so the Netflix stream will turn into a blocky mess for 10 seconds or so.
I also don't like the idea of games-as-a-service in the sense that I only own a license to the game for as long as I pay a monthly fee. I'm okay with licensing content within the game (i.e. DLC), but I shy away from subscription games services like Xbox Games Pass or EA's Origin subscription.
A streaming game product like this doesn't solve that issue. Additionally, the latency issue, like you pointed out, is too much for me to consider _not_ owning my own gaming hardware.
EDIT: A "Stadia store" was briefly mentioned shortly after my comment, it looked like it was tied into Google Play.
I'm guessing developers will be compensated with "exposure" :-)
I'm very curious about the expenses game developers will face too, which definitely wasn't brought up. Id Software bragged about just using one Stadia instance, but then later demonstrations showing split screen gaming "without compromise" was shown running each screen on a different Stadia instance. While Google suggests this makes split screen gaming possible again, my guess is that split screen gaming costs the developers twice as much to implement as a service cost, since it uses twice as much hardware. Making it probably mostly a nonstarter anyways.
Even for people that don't have gaming computers or consoles, you can put a decent one together that costs less than one of Google's phones
I want this to work.
This may result in a visual glitch, but you won't have missed your target.
Of course, the hardware you have to throw at it to do that quickly becomes ridiculous, unless you can come up with a way of efficiently rendering the multiple versions.
[1] https://www.libretro.com/index.php/retroarch-1-7-2%E2%80%8A-...
That said, I had my Bootcamped MBP with a XB1 controller connected over Bluetooth and any input lag was negligible with the Assassin's Creed Odyssey Project Stream demo.
No matter how good Google's CDN network is, you can't compete with sub-1ms in-network round trips. There's not really any clever "latency compensation" you can do here.
Even if you rewind state to apply input, which games probably won't support anyway, that still only helps you close the gap added by internet latency. A gap that in-home doesn't even have in the first place.
You still have an idle graphics card, but you can enjoy the benefits of it from any laptop in your home network. My gaming desktop is wired, my Linux laptop is wifi, and I get 4K @ 60FPS flawlessly.
However, I would prefer not to have to upgrade in the future and offloading the intensive parts to Google would be highly preferable.
Unless you have unlimited budget I find that buying cheaper cards more often will give you better average performance.
What makes you think games would just be included in Stadia?
Competing streaming services here are around $10-20/month without any games included. Origin's game access is $15/month without any game playing hardware included.
So... what will stradia end up being, and at what price? Would you pay $30/month for this? When a PS4/Xbox is only $300-400 in the first place, and has a lifespan of ~5 years?
Origin isn't relevant to this conversation.
$15/month for a streaming service + $15/month for game access service equals, drum roll, $30/month.
What are you not getting here? Origin is relevant as a reference point for what the games themselves cost on a subscription.
Stadia still needs to actually pay for games, after all. Either as a subscription (reference origin's pricing), or you straight up buy them at the normal $60/pop.
So instead of spending $1500 on that piece of gaming equipment spend $250-300 instead. That's what you're actually getting here. A very slow CPU with a mid-range GPU. It's a budget build that's not going to drive 4k@60fps with the bells and whistles. Not remotely close. It'll be fine for 1080p gaming, though, but that's realistically it.
They haven't fundamentally solved the latency issue, 50-100ms of ping is brutal and the people who could be a target for this generally won't have a fiber internet connection.
Network gameplay works via misdirection, predictable simulations and knowing how to converge a divergent state(including rewinding in time to re-play hittraces and the like). Streaming gets you none of that.
It'll work for high-latency sensitive singleplayer games but not seeing it scaling out wider.
Now you can program your clients and server like they’re all on a LAN. And now you can trust the client. (The client, not the inputs.) Those are huge reductions in complexity. This makes delivering good multiplayer easier, not harder.
But I do agree that input/display latency is going to be a thing. Results may be highly variable and you’ll probably know two people in the same city who have very different experiences with it.
At my day job I'm working on a AAA multiplayer game, and making a properly server-authoritative game with significant client-to-server latency mitigation imposes serious costs in dev time and complexity and security review process. And now, at least in theory, they mostly disappear.
The input/display lag problem is different from what we have now. It's new. We don't, of course, compensate for lag from input to client. (Between client and server, yes, but that's actually a different beast.) And yes, you won't be able to do much about it. You'll be relying on Google to keep your latencies down. They said they're deploying to 7500 nodes, so there's hope. Time will tell. I'm hopeful a lot of people will be reporting single-to-low-double-digit latencies.
But this is much, much more than just moving around the same problem.
I'm used to designing systems and games that work in 100ms+ latent environments since that's what you see in production(or try mobile where it can be 2-5x that).
We were doing 250ms ping times back in '99 with games like SubSpace and the like where you had a predictive game model/design. Yes you lose the overhead of designing for latency because it just won't fundamentally work in those scenarios.
This does, however, leave VR high and dry. Wonder if they have any ideas about that angle.
Wifi for instance is completely out. All it takes is someone turning on the wrong blender/micowave/power adapter and just the smallest bit of RF pollution will put you into unusable territory. Gets harder as your population density rises.
You can already see this with in-house streaming today. I can stream Netflix but my local LAN XBOne streaming is pretty unusable for things like Monster Hunter.
As your population density lowers the chance of you having the sub-30ms latency w/ 0% loss decreases significantly.
Like I said in the root, they'll able to get this to work for non-action(~200ms window) singleplayer titles, the idea that you could use this for multiplayer is laughable.
Will it be a worse experience than a 2080 Ti on your desktop? Sure. But will Stadia be good enough to win over the expense and hassle of the alternative? Maybe!
Regarding your last point, I'm not sure Id would stake their reputation on delivering a game that is going to be "laughable" on Stadia. They might know something we don't.
We'll see how it works out.
But what do you do when they don't? This can happen in multiplayer games with low population or during "bad times".
You might have less trouble because latency is more predictable, but you still need the whole rewind stuff.
That level of lag definitely makes some games implausible. (Part of why I bought a big-ass OLED is that newer TVs, but particularly OLEDs, seem to have generally consistent display lag; being able to play Tekken 7 on a 55" screen with only 21ms of lag is nice, but to do it on my old LCD is like playing underwater.) But I think it's going to be more practical for more games than the current conventional wisdom suggests.
When you're stuffing the pipe w/ 4k encoded content that puts a lot more pressure on the connection and added packet loss or delay is brutal in that case.
All of this stuff is subtle but it compounds in the same way that the average gamer won't be able to immediately spot 30 vs 60 fps but they'll know that the game "feels" better at 60.
Totally a concern. When OnLive closed people lost all their games, no refunds. While if you went to the store and bought a physical disk. It'd still be yours.
The average console quality game is about 50 or 60 bucks nowadays. If you are an active gamer, you could lockup and lose hundreds or even a thousand dollars when they decide to close it.
Most of Google's services are free, but paid ones I'd hope they wouldn't as easily abandon too.
I haven't bought a game on physical disk for a long time. My impression is that in many cases, you're still going to be downloading a big release day patch upon installation. And for multiplayer-heavy games, it doesn't matter if you have a physical disk when the servers are shut off.
Also Microsoft or Sony weren't companies known for gaming when they entered this space and now they are the leaders.
Sony entered at the right moment with the right tech, MS was at their peak when they started xbox.
Larry Land even started working on Azure solutions. They were more stable, however, more expensive.
With all that said, there's definitely an interest to be hardware light. So this move into a hosted service should probably be pioneered by the big three hosts: AWS, Azure and Google. They are the biggest consumers of hardware at the moment. And have a hell of a lot more financial backup then Onlive (who only had one product to sell). Xbox has but to partner with their Azure counterparts. Although, I grant you there's probably little incentive due to them being hardware sellers.
All this hardware could have been used for better purposes, of course. Like AI training or Pharma... but like in Stross' Accelerando - it's all going to be used to run economics 2.0 anyway.
To have the data of every choice and move in every second of every game would be quite useful to calculate the model of the particular user with the level of accuracy barely existed before.
Bungie had to build an infrastructure to push analytics from Halo into their system so they could get "heatmaps" of where players were in a multiplayer map. This fed back into their design process and was used to tweak maps after release.
Building something like that might be easier if all the state is living in a cloud server "right next to" the analytics pipeline, with no need to push state over the external connection. It could, hypothetically, even be built as a library Google supports directly.
- drastically lower power requirements
- no worries about hardware refreshes
- no waiting to download a 50gb game
- can play anywhere (steam link is getting there, but it's not as easy as opening a new chrome tab)
> hardcore PC gamers are never going to accept the latency
cool, they don't have to.
I would definitely try this on some quirky little games that I love to watch the streamers play, but too lazy to download. If the play could start in minutes, I would try it.
At this point I see comments like yours in here (from "games industry" insiders) and I see horse breeders decrying the car. "It takes gas instead of grass, it'll never work!"
Please.
Apparently it's a Goldman Sachs estimate. The last time that number was explicitly listed (in court filings) was $1B in 2014. The more apt metric would be what YouTube pays for original content, though I've only seen estimates/projections (a "few hundred million"):
https://www.engadget.com/2018/02/23/youtube-red-original-pro...
Why? What part of this is interesting?
Still a great thing for Linux and Vulkan. Too bad Sony refuses to support them on its PS consoles.
Plenty of extensions to choose from.
https://vulkan.gpuinfo.org/listextensions.php
Stadia's specs per Google:
Custom AMD GPU with HBM2 memory and 56 compute units capable of 10.7 teraflops
16 GB of RAM with up to 484 GB/s of performance
Radeon MI25's specs:
16GB of HBM2 memory at 484 GB/s with 64 compute units capable of 12.29 TFLOPs
12.29 * (56 / 64) = 10.7.
So this is almost certainly just MI25 chips with 1 or 2 bad 'cores'. Aka, the reject pile. From last generation, at that, since the current gen MI50 & MI60 are both 1TB/s HBM2 memory.
The multi-gpu part maaaaay be custom using the Instinct's extra interconnects instead of what used to be consumer Crossfire. But multi-gpu in consumer space is almost entirely dead, and nothing seems to suggest any developer would do anything for multi-gpu in this platform, either.
Much like web apps, YouTube, Netflix etc. today already work flawlessly on Linux, so will gaming too with this development. For most people, the browser is already the only "native" desktop application they use on their computers.
Web apps and web video finally made the Linux desktop viable, but also irrelevant in the bigger picture.
...sorry.
I wonder where the instances will be? Will they put them in their regions, or will it be rolled out to their CDN POPs?
I've used most of these applications in the past and I'm pretty salty they died. So yeah. I don't have a lot of faith in google keeping services it doesn't find profitable, my guess is the cost of compute will exceed the profit from selling games/game streaming/whatever and then, yes, they will shut it down. Don't act like its not something they don't do or wouldn't do. It's certainly not like making fun of the fat kid at the gym. Or are we turning into hail corporate over here?
Indie shops would probably struggle even more under this model.
The cost per user will vary drastically by the amount they play and how computationally demanding it is. I just don’t see it being feasible without personalized cost. Looking forward to paying minimum cost for minimum graphics too.
Microsoft/Sony are NOT going to sit back and let Google win here. Expect replies from them (especially with Microsoft's Azure).
The streaming services of gaming is coming.
https://blogs.microsoft.com/blog/2018/10/08/project-xcloud-g...
It's already failed multiple times. OnLive tried & failed. Nvidia's has been in beta for years. Sony has one.
Everyone in this area has tried this. Nobody has seen what could be described as "success", and the cost models so far have been ludicrous. Turns out renting Xeons and Radeon MI Instincts in a professionally staffed, maintained datacenter is way, way more expensive than a handful of consumer chips in a box in the living room with nobody on-call to monitor it.
The GPU here looks to be basically a slightly cut-down AMD MI25. That'd make a single GPU in this stradia cloud gaming service costs more than 10 xbox one x's. How do you make that price-competitive here?
I'm sure they got bulk/promotional pricing from AMD, plus they're very good at both running hardware with low overhead and packing it efficiently.
You can't really pack the hardware here since it's latency sensitive. It's straight dedicated resources to an array of VMs. Dedicated CPU cache, even, hence the odd 9.5MB L2+L3 number.
Bulk pricing only gets you so far here. You're still talking gear that's categorically way more expensive than similar performance consumer parts. Not to mention all the other costs in play - data center, power, IT staff, etc...
Making this price-competitive is a big problem
The other costs (power, people, etc.) are amortized over Google's array of services.
Last but not least, it would be very dumb of them not to run batch workloads on these machines when the gaming service is idle. I bet $1000 these puppies run under Borg.
Power doesn't really amortize, and neither does heat.
And capacity still had to increase for this. They didn't just find random GPUs under the table they forgot about, and now that they have a massive fleet of GPUs it's not suddenly going to start handling dremel queries.
This all still costs money. A shitload of it. Someone is going to pay that bill. More ads in YouTube won't really fund gaming sessions. So will this be ad breaks in the game? No way that's cost-effective for the resources used. Straight-subscription model? This seems most likely, but how much and how will you get people to pay for it?
I know from experience that Google is very cheap. You tell Urs you saved a million dollars and he'll ask you why you didn't save two. Or five.
If this takes off, the pricing of the service will pay for the hardware (assuming they did a reasonable job there of baking it in). Even if it doesn't, organic growth from other, much larger Google services can make use of the idle hardware.
For the record, I was involved in a couple of projects that required a lot of new hardware. One of them even ended up saving the company a lot of money in a very lucky, definitely unintended way.
This strikes me as rather amusing. Google was having such trouble getting their hands on enough GPUs that they decided to build custom hardware accelerators (TPUs) to fill the gaps.
I'm sure they'll find a use for these.
Could they be using these GPUs for other purposes during idle times (AI model training, cloud GPU instances)?
There's also an old paper by Murray Stokely and co. about the fake market that was created to make the most use of all hardware planetwide.
HBM2 memory is super expensive. Like, rumors are 16GB of HBM2 is $320 expensive. Toss in anything custom here and there's zero chance this is under $600/GPU.
Even in the hotly contested consumer market the 16gb HBM2 Radeon VII is $700. And that doesn't have any high speed interconnects to allow for sharing memory with CPU or multi-gpu.
>consoles
better join the PC Masterrace.
https://money.cnn.com/2015/05/08/technology/aol-dial-up/
Maybe I'm super negative here but I don't see any pros to this development. Today's games are already dumbed down and steered towards profits only. The only games I play these days are indie games made by 1 person up to a handful of people.
We really don't need more but we need better. (This applies to many other areas too)
Pay-per-hour would lead to shorter, more thoughtful games, rather than long grindy ones.
Similarly, monthly subscription where you get access to a list of games actually helps indie games. Being able to jump into an indie game you already own within seconds, compared to having to buy and install a small game you've never heard of before.
I'm not sure about that. Currently tons of people pay monthly for grindy MMO games. Expanding that sort of revenue scheme to single player games would encourage devs to create more Skinner boxes to keep players "engaged" over the long run.
Players aren't going to spend hours grinding if they know they're paying extra money for each of those hours.
Or it ties money directly to length, so you need your game to be 60+ hours to justify it's existence. Or people won't pay that much so we don't get any new dark souls or RDR2 length games.
Pay-per-hour means developers have an incentive to build addictive games that keep you just engaged enough to keep playing for an extended time.
> Similarly, monthly subscription where you get access to a list of games actually helps indie games. Being able to jump into an indie game you already own within seconds, compared to having to buy and install a small game you've never heard of before.
Monthly subscriptions mean that the platform has to choose what games to include and promote based on what is most likely to make users find value in the platform—which means focussing on those with widest appeal, unless their recommender engine can get enough signal to reliably predict niche interest.
Most games would be developed for Playstation + XBox + PC + Cloud service to maximise audience (unless one cloud provider gains a monopoly, something so far nobody managed to do in either game platforms or cloud computing).
Of course you can decide to develop only for the cloud platform of your choice, but that's nothing new. Microsoft and Sony already pay you good money to make your game exclusive to their platform (if you're lucky even if you're indy).
The biggest concern is going to be internet availability and data caps. I tried the beta for Project Stream (now Stadia) this past winter. It worked well and was impressive, but I have a good internet connection with a 1TB cap. I have friends with much worse speeds and harsher caps and I am not sure if this would be viable for them.
I am more concerned about 'physical' gaming being phased out. I doubt Stadia will do this soon, but I like building a new computer every 4 years for playing games. Maybes it's something I won't actually miss(like how I don't miss CDs for music), but it remains to be seen.
The trap that you should avoid falling into is assuming that's the only way games can thrive. F2P games are huge and dominate the conversation, but I think indie games are the best they have ever been. My favorite game last year was Into The Breach, and by all accounts it sold well in the same market that Fortnite dominated in.
Games are evolving in weird ways, but it's in a multifacted and diverse way.
Is the grind a way to slow progression in order to make micro-payments more desirable? Fuck. That.
I really enjoyed it for the first 20 hours or so.
Also, it doesn't have that high of a playerbase.
It is kind of another beast altogether for Google, Shadow, Microsoft (xCloud), etc. to dedicate actual hardware for your usage. Shadow and others like it are essentially remote VM w/ GPU that you rent by the hour. Stadia, xCloud, etc. we don't know what the business model is going to look like. The only positive here is that Microsoft, Google, and Amazon are all cloud infrastructure companies so in theory, they could price their offerings cheaper than a company that is a tenant on their systems.
Microsoft has made clear that they want Xbox to be a platform independent of the physical box they sell to people. They are 100% going in the direction of Stadia. The big question is how they are going to do it in regards to their next console.
That would be considered unfair competition and they would probably be fined by EU, I guess
The vast majority of your support department can be let go.
Indie developers don't get a cut of console sales, but they do care about how many people buy/play their games, and if this widens the market it seems like it'd be a win for them to me.
That’s just speculation mind you. Things like Stardew Valley would probably do even better.
Also, this will fail miserably if Google thinks it can continue with its current attitude toward user support, so there is hope.
So it could result in a renaissance of the classic AAA 3D single-player games, perhaps.
This is going to be an alternative for some types of games.
And quite on the contrary I think it would awesome for indies, since they tends to be pretty short. And it is much easier for them to gain consumers, if they went viral over streaming, many of which are, like horror games.
No, they didn't, and every single thing you've suggested about their gaming platform is also true about every other streaming platform, and yet none, literally none, of those platforms have tried any of the things you're complaining about.
You basically just made up a bunch of things to complain about because nothing in the actual release was objectionable. Yours is not a helpful way to react, my friend, though it is a popular one.
I’m open to debate about how this could be priced, but I’m pretty comfortable pointing to existing cloud computing business models or streaming services as a precedent.
I would invite you to come up with an alternative business model for serving people who like to play high end graphics games for many hours a day.
Why are you this comfortable? Netflix, YouTube TV, Twitch, Sling, Amazon Prime Video -- basically all streaming services offer flat rates, not per-MB ones.
Further, all existing game library services tout unlimited gaming as a primary selling point! That's the primary reason you opt into Gamefly or Nvidia Shield, at least according to their own marketing.
And this offering is not for high-end gamers. It's taking the benefits high-end gamers get for their investment into their hardware, and making it available to the millions of more casual gamers. This isn't for high-end gamers, so creating a business model for them using Stadia makes no sense.
Finally, you're not thinking of this at the right layer if you're thinking in terms of things like s3, ec2, lambda, etc.. This is the product that's built on top of those, and the single price problem has been present for hundreds of years. It's a solved one, just ask any current MMO or hell, any clothing manufacturer. You're basically saying that an XL t-shirt is going to cost the same as a S t-shirt, despite tens of thousands of examples to the contrary.
YouTube (and Google Play TV & Movies, which appears to carry the same for-sale/rent content in a different storefront) and Amazon Video also both offer purchase of individual content items as well as a common flat rate subscription to certain content.
I don't agree. Anything is scalable.
This does benefit from economies of scale, but it’s not something you can just solve with infrastructure and fixed cost. No matter how many computers you have, you’re still going to do multiple orders of magnitude more calculations to render high end games than stream a song. And you’re going to deal with difficult load balances because every twelve year old gets home from school at the same time (exaggeration but point stands). GPU time costs money.
GPU time costs money but it's a fixed cost, doesn't matter what the GPU time is being used for, therefore it won't be per-game. The end.
For similar reasons, you’re not just going to rake in the money mining bitcoin because you bought a bunch of computers.
Or maybe to make the point even stupider, you could make a game about training neural networks same as you would on a real cloud service provider. If you can understand why google doesn’t charge a simple monthly flat fee for cloud computing of neural nets, you can understand why they can’t charge a simple monthly fee for computing neural nets in a game.
Not sure how would that affect game developers though as there won't be any attractive way to programming for young people; before, making an own game was quite an attraction to jump into programming and a motivation to study hard.
These huge companies will keep throwing money at streaming because they see gold at the end of the rainbow, but I’ve seen no indication that they have a plan without giant question marks before the “profit” step. Look at YouTube, something Google acquired in 2006, is dominant in its space, and still doesn’t make them money. Yet somehow games, which are more finicky and reliant on universally good internet connections is going to work for them?
This is trend-chasing, once again without any real new ideas to overcome existing challenges.
It's apparently deeply tied to YouTube, and particularly to help provide an onramp for more (often ad-opportunity-generating) gaming content on YouTube a service that is both an ad-platform and the venue for at least two distinct revenue generating premium services (the older of which is ad free) that Google has not strangled in the crib.
If it runs for 4 years and makes money, that's a success for Google. It doesn't need to be permanent to be useful.
1. As subscription services take over, the upfront revenue game studios see will drop. This is just simple math: Xbox Game Pass costs $10/month, which means its the total cost of a AAA game over 6 months. In a traditional model, many gamers would expect to buy, lets say, 2 AAA games per year. In this new model, I can play as many as I want. And even if I only play 1 or 2 every year, I'm almost definitely going to be "dabbling" in the collection for other games I may want to play, ESPECIALLY if they're instant-on like Stadia. Even if they pay-out to studios based on some metric derived from time spent in game, there's no way studios will get the same level of income as they did before. (note: this is exactly why Spotify is having such a hard time, and why they're branching out beyond music. royalties abstracted behind a subscription service suck for the bottom line)
2. So upfront revenue drops. How do studios make that up? In-game transactions. They're already huge, and they'll just keep getting bigger.
3. So what, micro transactions (mtx) are the "new normal". Well, the top 5% of games can afford that decreased upfront revenue by making it up in mtx (think: Fortnite, Apex Legends, CoD). The trailing 95% can't (think: Indie titles).
4. Beyond that, you can bet your bottom dollar that Stadia will pay out tons to the AAA studios just to get their names on the platform, given that Google has no first party studios to speak of. Assassins Creed gets enough upfront revenue to make it worth their while, meanwhile the next indie darling is left out to dry, further balkanizing the gaming industry
5. Switching gears: A massive number of software engineers in the industry entered it because of gaming. Games, even back in the 90s, were such a clear application and value of computers that it was obvious, even to children, that they'd be something huge. It inspired a generation, to not only play, but to mod and even make their own. Now, we're moving that all off into the cloud, hidden from the next generation. Google wants you to own a Chromebook and consume their products, not understand how they work.
6. Speaking of modding: Its literally the source of the world's most popular games. Battle Royale? You can trace its roots back to mods for ARMA and Minecraft. MOBA? Dota, a mod for warcraft. Creativity happens in environments that large corporations can't recreate, and traditionally a great platform has been starting with a base game, a great game, that some studio created, then exerting your creativity on top of that platform. It benefited everyone, including large AAA studios who could then copy your idea and make millions. Yeah, good luck modding on a blackboxed server a hundred miles away.
7. But fine. I guess we're moving into the future and this is part of it. Except, there are millions, even BILLIONS, of people around the world without the internet capability to even join this service. Google tried to help solve this with Fiber, and gave up. Its fucking hard. They'd rather do easy, cool things, like cloud gaming. Modern consoles are bad enough; my brother, who lives just an hour outside of a top 10 US city, recently told me that he downloaded Fortnite on Xbox for the kids. It took a week of 24/7 downloading. Most games drop with day 1 patches in the dozens of gigabytes, even if you buy the disk in-stores. The sheer arrogance of Google, to get up on stage and claim this service is gaming for EVERYONE, the apex of accessibility, is disgusting to me. They're stuck so far up their own ass they've become the ouroboros.
8. Well, streaming games have taken over the world. Let's say you want to compete with Google on this streaming game front. All of the top three cloud providers now want to get into game streaming. So, no way can you compete on cost there with them; they own the data centers and give their game streaming div...
To that last point; it is impossible to understate how fundamentally important Nvidia's Pascal architecture has been to the development of both gaming and AI. In my mind, its the most important computing chipset invented in the 2010s, and belongs among the "world's greatest" chips next to the Intel Core architecture, Apple's A-series, Pentium, and the 8086. It put Nvidia, quite literally, 5 years ahead of the competition almost overnight; AMD is still catching up, three years later, to the perf-per-dollar and perf-per-watt of the GTX 1080.
That chipset, and the cards that were made with it (GTX 1080/1080Ti namely) were the first indication that DC rendering with a stream-to-client architecture was actually possible for video gaming. Before that it was hard to make an economic case for it.
OnLive was purchased by Sony, and its easy to conclude that they repurposed their tech for their Playstation Now service. So it lives on.
Also, to point 7. guess what billions of people can't afford a console or an expensive PC rig. Yet, in developing countries data is already very cheap if not free. So, this DEFINITELY is a big step closer to unlocking games for them.
By comparison, shipping "edge devices" (aka, uh, COMPUTERS) running an 845 Snapdragon or Tegra (like the Switch) is cheap and getting cheaper. What makes more sense: asking someone in a developing country to pay $200 one time for a general purpose computer useful for everything including 1080p gaming, or $10/month in addition to, uh guess what, some computing device they'd already have to own to access Stadia.
The adage goes: never underestimate the bandwidth of a station wagon full of tapes barrelling down the highway. I mean, the follow up is usually "but never forget the latency", though in this case that doesn't apply. Point being: the internet isn't the answer to everything, but if the only tool Google has ever known is a hammer then every problem is going to look like a nail.
Reminds me a bit of the 30/60 fps fights a few years ago. Sure, 30fps games look more "cinematic" and 60fps movies look uncanny, but 30fps games feel less responsive.
Google Cloud has regional US DCs in western Iowa and central South Carolina. A midpoint between those two locations roughly lands on Nashville TN, which is ~600 miles away from either. Light could make a roundtrip of that distance in 6ms. Of course, the internet doesn't allow for latency at the speed of light, but that's the physical limit, and that's plenty; a typical internet browser alone has input lag of 10ms [1]. In order to achieve 60fps, frames have 16ms to be rendered.
But the regional DC is only the worst-case, because they've said they're deploying these things in 7500 locations around the world. That's unprecedented scale for a tier 1 cloud provider at the edge. They know that they have to be close to consumer populations.
Also consider this: Once cloud streaming takes off, we're going to see deeper integration into the frameworks and game engines themselves. Imagine a game engine which is built for streaming. It could do input prediction, doing a "light rendering pass" of frames for N possible inputs the input buffer might receive on the next frame, before it receives them. These custom chips they use have plenty of headroom to do this at 1080p, and most controllers have, what, 12 buttons + all of the joystick states? Depending on the game this might be possible (example, hard to do in multiplayer). Combine that with the natural advantage a cloud-hosted multiplayer game would have in networking with other clients to resolve game-state, and you can see that its not just a strict downgrade; it might be possible that we'll see improvements in the performance of games beyond just the typical "new year better graphics" cycle.
[1] https://www.vsynctester.com/testing/mouse.html
[2] https://shadow.tech/
Time has shown that the best revenue model for these kinds of services is finding a monthly price, or multiple tiers of a monthly price, that adequately cover the costs incurred from heavy users and light users alike.
For example, Valve will keep updating/supporting/hosting events of CSGO even if it has <1Mil active users on PC, but I can totally see Google pulling plug on it and burn dedicated followers.
Bad example because CSGO has been sitting at #2/#3 on steam's active player charts for years and years. Consistently ~10x the player count of, say, Rainbow Six Siege. Continuing to invest resources into a game with CSGO's player count is just all around solid business strategy, particularly since it has ~500k concurrent users at any given point. Active player count would be way, way over 1Mil.
TF2 would be a better example, but I don't know how much (if anything?) Valve is still doing with the small TF2 community. It still has a respectable 50k concurrent user population, but has Valve been ignoring them? Are there still updates for it?
Google Books (14 years old), Google Finance (12), Google Scholar (14), Google Fonts (9), etc.
broken and abandoned http://sappingattention.blogspot.com/2019/02/how-badly-is-go...
>Google Finance
https://www.marketwatch.com/story/google-finance-as-you-know...
I saw the OnLive demo. It was very slick, and the concept seemed more than viable, it was inevitable.
I suspect that Google will have better infrastructure than OnLive, and that their customers will have better Internet.
Unfortunately for me at least, what I want is more akin to a cloud VM where they can host my games and I stream them wherever, which this isn't it. I've already got games between Steam, uPlay, Origin, etc. I have zero interest in buying a game 2x just to be able to have more portable play.
It'll probably be a mix of pay-once games, lots-of-games-for-a-subscription, and pay-once, play-until-Google-shuts-the-service-down games.
However, a gaming session means to have a dedicated daemon running for you somewhere. I doubt this can be deployed anywhere on the spot, but perhaps they have some amazing technology for that.
CDNs aren't helpful if every stream is unique.
It's not the edge cache itself that matters; it's that many ISPs peer directly with Google.
Google Cloud can route the audio/video/keyboard packets mostly over Google's private network and then only use the public internet once it gets to your ISP (or their transit provider). This provides Google with more control over how the packet gets to the end user.
Google provides a similar service to Google Cloud customers as the "Premium Network Service Tier".
https://cloud.google.com/network-tiers/docs/overview
https://en.wikipedia.org/wiki/Input_lag
It's the challenge of edge computing datacenter design.
EDIT: https://www.pcgamer.com/gaikai-founder-blames-onlives-pc-foc...
> the company had deployed thousands of servers that were sitting unused, and only ever had 1,600 concurrent users of the service worldwide
They were burning through all of their money because they highly overestimated the audience. That's one of the main problems the cloud was made to solve.
Traditional GPGPU customers, and machine learners, usually use CUDA instead of OpenCL / VulkanCompute / DirectCompute. At least from my position is looks that way. I’m a freelance developer and my clients are picking CUDA in ~75% cases. The rest is DirectCompute, if computing on client PCs and hardware compatibility is required.
Furthermore, there are other technological advances that have come even in the past 2 years that help facilitate this, like Secure Reliable Transport protocol (https://www.srtalliance.org/).
Things were just too stacked against OnLive at the time to make it work, but most of those barriers have since been removed after they had gone under.
Depreciation schedules for cloud services are one thing. Consumer products are another.
They lack the ability to commit to a space and cultivate it, casually changing course on a whim and leaving masses of people spinning in their wake. Just look at how they've managed YouTube and its guidelines, bans, etc. Or the Play Store and all of its debacles. Were I a game developer, I would never want to be beholden to a Google gaming platform.
Case in point: https://news.ycombinator.com/item?id=19432702
That's because states, entrenched providers and in some cases local governments made every inch of work cost a mile of time. The cost was making Fiber a non-starter... ISP's benefited from nationwide networks largely funded by governments, then used their monopoly influence to prevent any competition, and "hired" state governments to enforce their monopolies.
But we can hand-wave all of that away and claim it was Google's lack of commitment.
This is closer to their existing product portfolio, it doesn't have nearly as a difficult upstart cost relative to Fiber, and the business models are pretty clear. I think a closer analog would be the rumored reductions in their hardware division, but given that they're still making successful hardware + software, it could be more about reorganizing around priorities than it is an indication of their reduced investment
It’s not like a photo site where you can just download your content. Whatever is built to run on Stadia will not run on your home machine or anywhere else.
1. Release a product, make it decently good.
2. Kill it in a few years after it has gotten widely adopted.
3. Release an alternative product that does the same thing, but, a lot of times, worse than the original and/or with missing functionality from the old one that people actually care about.
4. Bonus step: make it really confusing (looking at you, Hangouts => Allo/Duo).
5. Repeat the cycle.
Notable examples of it that come to mind:
* Messaging: GChat => Hangouts => Allo/Duo => I got tired of keeping up with that they are pushing now
* Music streaming: Google Music => Google Play Music [just a rename, I think] => YouTube Remix => YouTube Music
* First party Android experience: Google Play Edition => [I feel like I am missing something in-between here] => Android One + Android Go
GPE -> Android Go does not fit the mold of the "cycle" you're talking about. Further, Android Go is substantially better than Android One.
Do you have any other examples?
As for others, I personally stopped using non-essential google products (i.e., the ones I believe that Google is unlikely to screw up without completely destroying faith in the company, like GMail or Maps), so I can't come up with anything. If you are curious, you can check out https://killedbygoogle.com/ and see if there is anything there that might fit the bill.
Disclaimer: I have zero affiliation with that website. I found it by trying to look up a list of killed off Google products, hoping it would bring up some memories. Then I realized that there are way too many products on the list for me to parse through, and the website has a nice and clean UI/UX that I am not ashamed to share here.
As for your bet, I'll take it. Let's say $100 USD?
Cloud gaming needs a killer multiplayer game that requires massive parallel compute (thus it can't be run locally) to make a case for sticking around.
Even if MMOs shut down, there is usually an opportunity to write a new server by reverse engineering the client. I can't really support anything about game streaming unless it was tied into an actual game you could install or play without relying on their servers to exist.
https://news.ycombinator.com/item?id=19429388
There is definitely some cool work here though: being able to have access to a library of games without lugging around a console or computer is an interesting idea with both pros and cons. The pro that I can think of is not having to be responsible for upgrades. The con is that we continue to move towards a world where we rent more and own less.