Google Stadia specifications and latency details
We finally have an idea of Google Stadia's specifications, latency and performance.
During today’s Google Keynote at GDC 2019, Google revealed Google Stadia, a new streaming service that will make it much easier for gamers to enjoy new, high quality games without a powerhouse of a computer. While the original presentation left us with a lot of questions, more about the game streaming technology is finally coming out. Here’s what we know so far.
Google Stadia specifications and latency details
What makes Google Stadia so intriguing is so much of what Google is promising will be available with the tech. Not only will it allow game streaming on devices like Chromebooks, cellphones, or any other Chrome-enabled device, but Stadia will also support cross-platform multiplayer and several other really nice features like seamless play between devices.
On top of all the features listed above, Google is also aiming to deliver gaming at 4K quality at 60 frames per second. That’s a bold claim to make when it comes to streaming video games, and it isn’t an especially easy one to pull off. Thankfully, Google has some heavy hardware under the hood here. Google Stadia will be powered by the following specifications:
- Custom x86 processor clocked at 2.7GHz w/ AVX2 SIMD and 9.5MB of L2+L3 cache
- Custom AMD GPU w/ HBM2 memory, 56 compute units, and 10.7TFLOPs
- 16GB of RAM (shared between CPU and GPU), up to 484GB/s of bandwidth
- SSD cloud storage
While the specs above are definitely something worth boasting about, what’s even more appealing is the fact that developers will be able to stack more than one together, allowing them to virtually make a super computer to run their creations. Digital Foundry recently had a chance to sit down with Google Stadia, where they were able to get some reads on the system’s overall performance.
According to Digital Foundry, they were able to measure Google Stadia at roughly 166.67ms while using a Pixelbook connected to wifi. Of course, this was all taken with the latest version of Stadia available right now, so things might change a little over the coming months as we move closer to Stadia’s full release to consumer.
It should be noted, though, that this was not a test controlled by Digital Foundry. Which means the results might be a bit skewed. Still, though, it’s something worth investigating if you’re interested in seeing how Stadia stands up to current-gen consoles, especially given the boasts and claims that Google has made.
We’ll update this article as more solidified information becomes available for Google Stadia's specifcations. For now, be sure to check out the rest of our GDC 2019 coverage for more of the latest news from the conference.
-
Josh Hawkins posted a new article, Google Stadia specifications and latency details
-
-
Those numbers make baby Jesus (and John Carmack) cry. Here's a great article from Carmack waaaaaaaaaay back about the crusade to minimize input lag:
https://danluu.com/latency-mitigation/
Carmack totally saw this coming, too... Damn he's smart.-
Many people play games on TVs with so much processing lag that the pixels might as well have been coming from a datacenter. Not everyone knows or bothers to enable game mode (sadly).
https://twitter.com/ID_AA_Carmack/status/1108145247224324096
-
-
So for those with Gaming PC's, or even the higher-end consoles, it seems like the benefits will be these:
1. Being able to try/play games instantly.
2. All the stuff about sharing save states over YouTube is neat, and easily accessible multiplayer through YouTube as well. This goes well with point one.
3. The potential gaming experiences that could be made when the rendering is done by a machine more powerful than the average consumer. They said "imagine a 1000-person Battle Royale". You could do thousands, or even tens or hundreds of thousands of well-optimized AI routines within an open-world. Or as they also demonstrated, make a 100% fully destructible city. The kind of stuff that is genuinely hard to do on consumer hardware.
For everything else, video bitrate will always be a huge roadblock. Things just aren't going to look as good on a big screen, or if you're sitting close to a monitor. I tried Project Stream and while it was neat, I'd never want to play a game like that when I could just locally render it. Of course this is a problem that will eventually get more and more mitigated by better data networking.
That'll be the point, especially combined with the three above points, when this kind of closed-loop platform could be genuinely competitive to even gaming PCs. We aren't there yet but now it's pretty clearly on the horizon. Hug your GPU's, everyone!-
#3 is a long way off given that it's going to require Stadia to have significant adoption to justify an exclusive that does this. Crackdown 3 also looks like a cautionary tale there although I suspect it was probably just before its time and the execution of the rest of the game suggests maybe the problem wasn't intrinsic to the strategy of cloud powered gameplay.
-
"going to require Stadia to have significant adoption to justify an exclusive that does this"
For any other company, yes. But Google has the sheer BANK to fund multiple hugely expensive exclusives just under the guise of "increasing user engagement". Hell, they could be super aggressive about it and make these first massively expensive premium products free. -
-
Crackdown 3 was weird because the whole game runs locally, how do you outsource just the physics? It's different if everyone is using this service. But I don't see many big exclusives happening any time soon, it just excludes so many people. I would be curious to see an MMO built on this service though.
-
I think more significant will be the cost implications. This whole model provides that the cost of rendering will be borne by the cloud, but that doesn't come for free. I haven't seen anything about how Google will bill the publishers for this, but I assume that they will. This feels like a more complex problem in the vein of what we saw with dedicated servers for online games too.. Will the publisher be responsible for that cost and maintenance?
-
-
-
I really don't feel that's the case. We are all in a bubble here so I can see why a lot of people who are heavily into PC gaming don't have much interest in the project, but think about all the people out there who want to play the latest games but don't want to drop $500 on a console or well over a thousand on a gaming PC. That's a huge untapped Market.
-
-
-
-
For some games I'm sure it's 'fine', but just imagine this: if you were using your computer, and the mouse movement lagged by about a tenth of a second for every movement, how acceptable would you find that?
I'm pretty sure the shack is the kind of place where people couldn't stand the feel of older generation bluetooth mice.
-
Its very genre dependent and the reason they compare it to the console port is that it runs at 30 fps.
That’s also why a game like devil may cry 5 doesn’t runs at 30 fps. And people like me have been vocal about games like gta4 having horrific latency, and it goes back to streetfighter on the dreamcast (the fighting community has been shittalking sf5 in particular, some games in the genre even mocking it with a sf5-mode adding latency). -
Here an old digital foundry article excerpt: * The lowest latencies a video game can have is 50ms (three frames) - the PS3 XMB runs at this rate, but few games reach it.
* Most 60FPS games have a 66.67ms latency - Ridge Racer 7, for example.
* 30FPS games have a minimum potential lag of 100ms, but many exceed this.
* Game developers should test their own games using the camera technique in order to weed out bugs - West says that Heavenly Sword's response slows down to 300ms just by turning the character, and reckons it's a technical issue that should have been resolved before going gold with the game.
* Citing GTAIV as an example, West suggests that a 166ms response is where gamers notice controller lag, which could also explain the Killzone 2 furore too.
* Game reviewers should accurately measure latency for their reviews where controller lag is an issue, in the hope that sloppy game response times come under far more scrutiny.
Game Latency Measurement
Burnout Paradise 67ms
BioShock (frame-locked) 133ms
BioShock (unlocked) as low as 67ms
Call of Duty 4: Modern Warfare 67ms-84ms
Call of Duty: World at War 67ms-100ms
Call of Juarez: Bound in Blood 100ms
Forza Motorsport 2 67ms
Geometry Wars 2 67ms
Guitar Hero: Aerosmith 67ms
Grand Theft Auto IV 133ms-200ms
Halo 3 100ms-150ms
Left 4 Dead 100ms-133ms
LEGO Batman 133ms
Mirror's Edge 133ms
Street Fighter IV 67ms
Soul Calibur IV 67ms-84ms
Unreal Tournament 3 100ms-133ms
X-Men Origins: Wolverine 133ms -
And for good measure more current examples, see if you notice a pattern:
Current-Gen Tests Input Lag Target Frame-Rate
Call of Duty: Infinite Warfare 39.3ms 60fps
Call of Duty: Modern Warfare Remastered 40.3ms 60fps
Battlefield 1 56.1ms 60fps
Halo 5 63.0ms 60fps
Battlefield 4 63.7ms 60fps
Titanfall 2 71.8ms 60fps
Overwatch 76.8ms 60fps
Doom 2016 86.8ms 60fps
Killzone Shadowfall Multiplayer 89.8ms 60fps
Killzone Shadowfall Single-Player 110.0ms 30fps -
-
Yeah, something seems off about that. I don't use my XB1 much anymore, but when I did, there was no discernable latency and I could play things like Rayman with split second timing without issue. Whereas using my Steam Link or Nvidia Shield which has a supposed ~120ms latency, those sorts of games are unplayable due to their requirement on instant reaction.
This whole thing has me confused, but I can say with some degree of certainty that the input lag will be the dealbreaker on even the highest of high end connections with Stadia.
-
I played Odyssey on Project Stream. The lag was there but it was perfectly playable. And I genuinely mean perfectly playable, I intentionally put myself on the hardest difficulty to test it. I had a quick reaction dodge that I was astounded the game registered in time. I don't know what kind of black magic prediction they are doing in addition to their optimized piping but it works.
I wouldn't want to play it because of the video bitrate, but the latency was really just not much of an issue. -
If PlayStation Now and OnLive are anything to go by, the real way you'll know that this technology is a failure is not that a bunch of hardcore gamers hate on however many nanoseconds of lag there are, it'll be because the tech and the implementation are basically ignored by everyone.
I mean think about it - a version of this exact idea that Google and next Microsoft is rolling out has been in Sony hardware for years now and no one fucking cares about it. There could be other reasons but basically this seemingly holy grail of streaming gaming has been in the hands of the general public for years and people aren't saying "well the tech isn't there yet" or "maybe when the pricing gets better" because if they did then they'd be talking about it at all, which they aren't.
And you can say OnLive was some outlier because they were a startup with a tyrannical leader that doomed the company but it's not like Sony is a slouch. They had this built into their TVs and they removed it because no one used it. They added it to the PlayStation 3 and removed it later and no one cared - and people sued when they took away the ability to run Linux from the PS3.
Maybe it really does need Microsoft money. Or Google money. Or maybe it's impossible no matter how much money you throw at it.
But maybe these things will fail because no one cares. -
-
-