Vapourware Google Stadia - "a game streaming service for everyone"

[J_C]

How could a 100-150ms input-lag cloud gaming service actually work. Who would want to play on a machine hundreds of miles apart rather than just using a console.

It's not 100-150 ms. Geforce Now (which I actually tried) has between 20-50 ms, based on your connection. At this moment my Geforce Now connection has 35 ms latency. It's perfectly managable if you don't play competitive games. If my country had a datacenter, it would probably go down to 20 ms, which is only slightly worse than a wireless mouse.

I don't actually use game streaming, because the games I play run on my PC, but I perfectly see someone how does not want to buy a PC use it.

 

[francisbaud]

How could a 100-150ms input-lag cloud gaming service actually work. Who would want to play on a machine hundreds of miles apart rather than just using a console.

It's not 100-150 ms. Geforce Now (which I actually tried) has between 20-50 ms, based on your connection. At this moment mine has 35 ms. It's perfectly managable if you don't play competitive games.

I really doubt the time it takes to send the information off to the cloud service, to process this information and then send it back would take 20-50ms. That seems impossible, unless you live next to the cloud service facility.

Here's an example of what we can expect (from a IGN article)

Input latency stats (in milliseconds)
Game------------Local PC 1080p--------Stadia PC 1080p-----------Stadia TV 1080p----------Stadia TV 4K
Destiny 2-----------------83-----------------------------150-------------------------------163----------------------------167
Shadow of the
Tomb Raider-------------63-----------------------------125-------------------------------213----------------------------279

 

[J_C]

How could a 100-150ms input-lag cloud gaming service actually work. Who would want to play on a machine hundreds of miles apart rather than just using a console.

It's not 100-150 ms. Geforce Now (which I actually tried) has between 20-50 ms, based on your connection. At this moment mine has 35 ms. It's perfectly managable if you don't play competitive games.

I really doubt the time it takes to send the information off to the cloud service, the time it takes to process this information and send it back would take 20-50ms. That seems impossible.

Here's an example of what we can expect (from a IGN article)

Input latency stats (in milliseconds)
Game Local PC 1080p Stadia PC 1080p Stadia TV 1080p Stadia TV 4K
Destiny 2 83 150 163 167
Shadow of the
Tomb Raider 63 125 213 279

Look, I don't have the equipment to test this, I'm just saying what the connection test on Geforce Now shows me. And when I tried an FPS on it, it was managable. It was noticable, but it wasn't something you can't get used to. Hell, people people play games in sub 30 fps, which looks awful, but they get used to it.

By the way, that IGN article probably is a load of bullshit. There have been tests on video, which show that a wireless mouse has a latency of 15-20 ms. And they say that it is 63-83? Nope.

 

[rusty_shackleford]

I had sub 30ms latency(iirc) when I was testing stadia. Somewhat noticeable in FPS(Anthem 2), not really noticeable in anything else.

 

[J1M]

You can't "Patch" the speed of light.

They were contemplating "predictive input" as a feature. Essentially pushing the buttons for you to reduce perceived latency at the cost of accuracy.

 

[baud]

Weird how they went about this. I'd always assumed that if one were to start out with cloud based gaming, if you started small and grew it over time eventually the technology necessary make this format more viable would arrive at some point in the near future - that is, emerging tech in reducing latency or a custom thin client the integrates with the service to improve performance. You keep the service going to continually improve. Think Google just spent crazy Google money with little vision. Reminds me of Microsoft moving into mobile phones for some reason.

The big corps have the money to gamble on these sorts of projects. If they fail, they lose a bit of pocket money, but a big success could propel them into the stratosphere.

Google in particular is pretty infamous for the amount of failed projects they back and develop in the hope of one being successful. In fairness, it seems to be a valid strategy.

Google's strategy of going for moonshot might be valid, but it looks like it haven't gotten anywhere yet (apart from Waymo, maybe?)

 

[Ravielsk]

Starting small would actually be very simple. They could start small by focusing on single player games, or non-graphics intensive games.

The scope or graphical intensity of a game have nothing to do with it. You cannot realistically split up a GPU between multiple applications in any way that would matter. Essentially it could help to service a few more people but it would only post-pone the problem for later when bigger games started hoping onto the platform. Eventually streaming has to tackle more demanding processes and there these sorts of work-around stop working either way.
Besides from a market perspective I highly doubt many people would be jumping to yet another service for a bunch of AA/indie games

But if cloud based FPS games require the same amount of bandwidth as a YouTube, they should have never stepped into this.

It actually requires more. I do not remember the exact numbers but a 1080p 60fps clocks around 20-40GB per hour or two. It strongly depends on the game. Naturally turn based titles or games with slower action and fewer colors take up less bandwith but at high resolutions and framerates the file size is high regardless.

Your vision...

Its not my vision that is how it works currently. Games and any other highly demanding applications are designed to run on dedicated hardware, not to run in some bizzaro multi instance server setup that most "cloud" applications use. So the only thing that can be done with that is to virtually partition the server hardware to perform like a local dedicated PC and send the results over the net. It could be theoretically made more efficient by redoing the code to run multiple instances of itself on the same hardware but really those are optimization pennies not magic that would let you turn 270 cores into 2700 cores.

It would also be very odd to renderer the same map 32 times, one for each customer. Essentially, the nature of FPS games would need to be changed to make Google's strategy work. Maybe that's why the huge outlay for vendors from Google. That is to say - right now in World of Warcraft, the client hosts the bulk of the heavy lifting. Once one connects to battlenet - Blizzards server maintains player interactions and such, but doesn't host the image of everyone else - your client does. FPS games are designed around limiting the work the server has to do. Creating a server with 32 video processors for 32 players sounds downright horrible. Why would the same map need to be rendered 32 times unless it was a limitation of the game itself? A multi-player game designed around thin clients would need to be built from the ground up, and keep the amount of data transferred as skinny as possible because Google can't control end user data connections, but will have to deal with support calls from everyone on DSL or such.

Its not strange, its how it works right now. You could potentially redo the games to have one set of GPUs and CPUs focused on rendering the whole map without culling and another set dedicated to running player actions and interactions but that alone introduces new problems where you would basically have to cut out a chunk of the available HW from the rest and make it exclusively process that one game. Which depending on the number of players and demand for other games might not be possible.

 

[rusty_shackleford]

There was a paper published last year that analyzed Stadia network traffic if you want to know more: https://arxiv.org/pdf/2009.09786.pdf
It uses about 10 GiB an hour to stream a 1080p 3D action game(rise of the tomb raider,) and they had a 10ms RTT.

In all cases, the 95 % percentile of the RTT values is lower than the duration of a single video frame (16.67ms), meaning that in these tests Stadia had the opportunity to interact with the player’s actions without any perceptible delay.

 

[Daemongar]

Its not strange, its how it works right now.

Not going to make a long post. I understand how it works now, and to bring it full circle, if their vision was to do it "how it works now" then it was a bad idea. The only way this will work is if the games themselves are rewritten as client server applications, which in turn factor in the efficiencies of client server technology. It's agreed that the Stadia games run shittier than local consoles, as they did nothing but introduce long distances between your controller and the the action. I expect more from Google - as described it was doomed from the start.

Also, they could start small - they could market their service to females. Females play mostly casual single player games.

 

[rusty_shackleford]

Its not strange, its how it works right now.

Not going to make a long post. I understand how it works now, and to bring it full circle, if their vision was to do it "how it works now" then it was a bad idea. The only way this will work is if the games themselves are rewritten as client server applications, which in turn factor in the efficiencies of client server technology. It's agreed that the Stadia games run shittier than local consoles, as they did nothing but introduce long distances between your controller and the the action. I expect more from Google - as described it was doomed from the start.

Also, they could start small - they could market their service to females. Females play mostly casual single player games.

But why would women use it? Most casual games can be played on cheap hardware.
The problem with Stadia was always content. When I found out users are expected to buy each title individually I knew it was dead. I originally thought it was a subscription service like Microsoft's gamepass, and it came with a bunch of games. Having to buy full-priced games that are completely locked to Stadia's library was just icing on the shit sundae.

The only games this service is useful for are AAA hardware intensive games. Coincidentally, these games also tend to cost the most. If you can afford to buy more than a couple of these, you probably already have or can afford your own gaming PC and don't need the service.

 

[Chinese Jetpilot]

Tried a few of the game streaming services, but settled on a service called Shadow; I'm located very close to one of their datacenters, so have about ~10ms latency. The best thing about it is is the ability to bring your own library.

 

[tritosine2k]

Its not strange, its how it works right now.

...

The only games this service is useful for are AAA hardware intensive games. Coincidentally, these games also tend to cost the most. If you can afford to buy more than a couple of these, you probably already have or can afford your own gaming PC and don't need the service.

Naive, compression cannot handle rasterization imperfections and perfect AA is only an option for artsy games.

 

[rusty_shackleford]

Naive, compression cannot handle rasterization imperfections and perfect AA is only an option for artsy games.

have no idea what you're trying to say here tbh

 

[tritosine2k]

Naive, compression cannot handle rasterization imperfections and perfect AA is only an option for artsy games.

have no idea what you're trying to say here tbh

https://www.gamesindustry.biz/artic...on-suit-takes-aim-at-google-stadias-4k-claims
The same thing the 4k lawsuit is not without merit because encoders cannot even handle it, and when it has imperfect AA the issue is compounded.

 

[Decado]

I've told this story on the Codex before but I met with a Stadia exec back in March of 2019 at GDC, as part of my job at the time. They were unbelievably arrogant, and even for Silicon Valley types who think they are smarter than everyone. I deal with those people a lot -- this woman was on another level, just supremely arrogant. We started telling her about some of the current issues in the video game landscape that we were seeing and she interrupted us and was like "We know all of this already."

Apparently, they didn't.

Anyway, I'm never glad when something fails but seriously, if any company had this coming it is fucking Google. Fuck them and their attitude. People tried to help them and they sneered at it. I hope Phil Harrison catches his penis in his zipper and cuts his dick, and he has a little cut on his dick for a few days and it is mildly uncomfortable.

 

[Keyboard Vomit]

I've told this story on the Codex before but I met with a Stadia exec back in March of 2019 at GDC, as part of my job at the time. They were unbelievably arrogant, and even for Silicon Valley types who think they are smarter than everyone. I deal with those people a lot -- this woman was on another level, just supremely arrogant. We started telling her about some of the current issues in the video game landscape that we were seeing and she interrupted us and was like "We know all of this already."

You sound like an incel that has a problem with strong women.

 

[abija]

Not going to make a long post. I understand how it works now, and to bring it full circle, if their vision was to do it "how it works now" then it was a bad idea. The only way this will work is if the games themselves are rewritten as client server applications, which in turn factor in the efficiencies of client server technology.

Their whole point is to play on any device which means client needs to only get input and present output, nothing else. There's no magic that can happen, there's a hard limit: you cannot play anything fast/competitive. Plenty of games to sell though.

Real issue is the target audience. Rich gamers that travel a lot and only in areas with great connectivity are a niche at best.
To move people to streaming they need to make it way cheaper for a long period of time or hardcore exclusivity for a long period of time.

Also, they could start small - they could market their service to females. Females play mostly casual single player games.

Small is relative. I thought they started small, for Google.

 

[Daemongar]

Not going to make a long post. I understand how it works now, and to bring it full circle, if their vision was to do it "how it works now" then it was a bad idea. The only way this will work is if the games themselves are rewritten as client server applications, which in turn factor in the efficiencies of client server technology.

Their whole point is to play on any device which means client needs to only get input and present output, nothing else. There's no magic that can happen, there's a hard limit: you cannot play anything fast/competitive. Plenty of games to sell though.

Would prefer to keep this abstract, but whatever.

• Build technology changes into your plan. There is a hard limit now BUT technology changes can catch up and be part of your business plan, if you are a technology company with a multi-year plan. Nobody expects the IT landscape to remain static. So, that's why I said "start small but stay in it for the long haul."
• Reduce the amount of data transferred. Other changes, like remodeling game structure should be part of their strategy to, at a minimum, reduce the amount of data transferred so that the burden is on the server, not the client, even for single player games. Stadia does have an SDK it appears - so I don't know how far along they are in this, or if its even a consideration.
• Improve online response. Changes such as adopting a steam-like system of matching players up with the servers with the lowest latency can be adopted, and they could change aspects of games from asynchronous to synchronous communications, or something as simple as sending subscribers some kind of ethernet thingamabob with game controllers that creates a dedicated 5Ghz wireless network for and only for the controller. The dumb thing in Googles plan is that Stadia is dependent too much on folks internet speed and home wireless network setup - if either is operating poorly, Stadia looks bad in addition to geographical impediments to communications.

Small is relative. I thought they started small, for Google.

Small as in - understanding the current barriers to a polished product, focus on what your service does best, don't highlight what it does worst in your advertising. My platform does Super Mario Cart well, but Destiny 2 plays poorly. Yes, you can gather more interest placing Destiny 2 in your add, but people are going to show up for Destiny 2, be disappointed and leave.

 

[Decado]

I've told this story on the Codex before but I met with a Stadia exec back in March of 2019 at GDC, as part of my job at the time. They were unbelievably arrogant, and even for Silicon Valley types who think they are smarter than everyone. I deal with those people a lot -- this woman was on another level, just supremely arrogant. We started telling her about some of the current issues in the video game landscape that we were seeing and she interrupted us and was like "We know all of this already."

You sound like an incel that has a problem with strong women.

The best way to get over this is marry a good looking and intelligent woman and fuck her brains out every night.

 

[Average Manatee]

I've told this story on the Codex before but I met with a Stadia exec back in March of 2019 at GDC, as part of my job at the time. They were unbelievably arrogant, and even for Silicon Valley types who think they are smarter than everyone. I deal with those people a lot -- this woman was on another level, just supremely arrogant. We started telling her about some of the current issues in the video game landscape that we were seeing and she interrupted us and was like "We know all of this already."

Apparently, they didn't.

Anyway, I'm never glad when something fails but seriously, if any company had this coming it is fucking Google. Fuck them and their attitude. People tried to help them and they sneered at it. I hope Phil Harrison catches his penis in his zipper and cuts his dick, and he has a little cut on his dick for a few days and it is mildly uncomfortable.

This is exactly the impression I got for everything about their launch. Fantastic, incredible hubris brought on by the expectation that being google and having infinite money means you can't lose. A bunch of mid-level managers who alternate between getting their cock sucked by their employees and the press but have no actual vision or experience got promoted to head a hot new project that required effort, but they put in none and instead assumed they were too big to fail.

 

[rusty_shackleford]

just realized I can still play destiny 2 for free on stadia, gave it a try again
I have absolutely zero detectable input lag

definitely do get stuttering when quickly moving my camera though

[edit]
Also decided to give GeForce Now a whirl after being told it has a beta chromium client now. About the same latency(that is, not noticeable in an FPS,) but no stuttering when moving my camera like in Stadia, the picture quality is much better. I honestly can't tell it from native except occasional stuttering.
Considering GeForce Now is bring your own library, it's by far the better service imo. Kinda crazy how much Nvidia outdid Google at this considering the massive advantages Google has.

No idea how the GeForce now desktop client compares, there isn't one for Linux.

 

[Decado]

Plus, GeForce was doing this shit back in 2014! I bought a Shield Tablet for this exact purpose and it was great. I actually beat Ornstein and Smough for the first time on my Shield tablet in my office, streamed from my home computer 45 miles away on a tuesday morning before anyone else was at work and I had the internet all to myself. Was pretty fucking cool.

It was never about the technology. It was -- and is, and always will be -- about the games. It's that simple.

 

[abija]

If you don't notice the input lag in a fps at 45ms latency to Stadia/GFNow you should check a medic. (or stop exaggerating for effect)

 

[rusty_shackleford]

If you don't notice the input lag in a fps at 45ms latency to Stadia/GFNow you should check a medic. (or stop exaggerating for effect)

My roundtrip delay on GeForce Now is <25ms, meaning it takes about 10-12ms for my input to reach the server.

 

[abija]

Same thing applies, even at 10ms. Fire something like quake champions, if it's still free. You can't track, you can't flick. It's worse than first generation lcds.