Relevant table from article: . Google Stadia Google Stadia 15mbps Project Stream PC 30fps PC 60fps Xbox One X Latency 166ms 188ms 179ms 112ms 79ms 145ms Latency (inc display lag) 166ms 188ms 200ms...
What is this latency based on? Is this a latency to get a frame on screen? Because these numbers don't make sense. According to the table "Display" lag doesn't exist for the Stadia which makes no...
What is this latency based on? Is this a latency to get a frame on screen? Because these numbers don't make sense.
According to the table "Display" lag doesn't exist for the Stadia which makes no sense.
If you have a 60fps PC game, on a 6ms latency monitor you're going to have 1/60 = 17ms + 6ms = 23ms of latency.
So where's the 100-23 = 77ms of lag coming from?
Also what's going to happen with online games? So you have this 166ms latency which I assume the latency to get a frame on screen, then you'll have 100ms of network lag in a multiplayer game so you could end up with over 250ms of lag.
EDIT: I figured out they were using the latency of a PC over WiFi to another device, which I personally notice the latency.
Unfortunately, I don't think that will ever be possible. VR is extremely latency-sensitive, and you will get sick if the view doesn't move correctly and immediately when you look around. I'm quite...
Unfortunately, I don't think that will ever be possible. VR is extremely latency-sensitive, and you will get sick if the view doesn't move correctly and immediately when you look around. I'm quite accustomed to VR at this point and regularly play games considered "high intensity" that make others feel motion-sick, but anything that lags/hitches even briefly while I'm turning my head will still make me start feeling sick almost instantly.
How much latency is too much? Less than you might think. For reference, games generally have latency from mouse movement to screen update of 50 ms or higher (sometimes much higher), although I’ve seen numbers as low as about 30 ms for graphically simple games running with tearing (that is, with vsync off). In contrast, I can tell you from personal experience that more than 20 ms is too much for VR and especially AR, but research indicates that 15 ms might be the threshold, or even 7 ms.
Since cloud gaming requires a round trip (you send your input to the server, it processes it and sends back the result), that means you actually only have half that long in each direction, and it's not possible to send data over the internet anywhere close to that quickly.
To add: The tested setup was 1080p30fps on a 200mbps connection. Extrapolate a little, and that's ridiculous amounts of bandwidth required for the refresh rates and resolutions that a good VR...
To add: The tested setup was 1080p30fps on a 200mbps connection. Extrapolate a little, and that's ridiculous amounts of bandwidth required for the refresh rates and resolutions that a good VR experience requires. Even if you completely discount the latency, just getting the bandwidth to do anything VR-like will be very difficult for most users.
That really depends on your connection and the location of the servers you're connecting to... my buddy in Virginia has FIOS and regularly gets pings of <20ms on the East Coast servers for Rocket...
and it's not possible to send data over the internet anywhere close to that quickly.
That really depends on your connection and the location of the servers you're connecting to... my buddy in Virginia has FIOS and regularly gets pings of <20ms on the East Coast servers for Rocket League. We often joke that the servers are in his basement. But yeah, overall 20ms is unlikely to be achievable for the vast majority of people in most cases, at least for the foreseeable future.
That's also pure network ping too though, without the other pieces of latency (input, the game processing the input, the display processing the stream data, etc.) and at some point you even start...
That's also pure network ping too though, without the other pieces of latency (input, the game processing the input, the display processing the stream data, etc.) and at some point you even start running into the speed of light as a limit. For example, let's say that you're allowing 5ms of network latency in each direction, and let's give 1ms of that for time spent going through routers and such on the way. That leaves 4ms for traveling through the physical cables, and if you use 124,188 miles/sec as the speed of light in fiber, that means the server you're communicating with can not be more than 497 miles (800 km) of cable away from you.
For the other pieces, a game running at a constant 100 frames/sec is spending 10ms on each frame, but even that is probably too much if you need latency this low. So you need a server capable of running the game at 200 FPS or so within 500 miles of every player, and even that's already pushing 15ms without including multiple other sources of latency.
It might be theoretically possible, but very, very unlikely to ever happen in realistic conditions.
That's not to rule out streaming technologies completely. A dual systems where the scene (or parts of the scene) are rendered locally, but much of the heavy lifting ("game logic") is handled...
That's not to rule out streaming technologies completely. A dual systems where the scene (or parts of the scene) are rendered locally, but much of the heavy lifting ("game logic") is handled server-side is still possible. It might rule out complex lighting or other expensive operations, but I can see a future there.
Think a more advanced version of Oculus' asynchronous timewarp.
Yeah that is what I was thinking. Do light computing on a phone like device and then do ray tracing or something super compute heavy in the cloud and bake that in or something.
Yeah that is what I was thinking. Do light computing on a phone like device and then do ray tracing or something super compute heavy in the cloud and bake that in or something.
What about a sort of asynchronous spacewarp along with cloud rendering? Have a local computer render the old frame and movement data and have a cloud computer render the graphically intense frames?
What about a sort of asynchronous spacewarp along with cloud rendering? Have a local computer render the old frame and movement data and have a cloud computer render the graphically intense frames?
Maybe, but that's getting pretty far away from a game-agnostic streaming system and into custom client software for each game, where you have part of the game running on the local device and part...
Maybe, but that's getting pretty far away from a game-agnostic streaming system and into custom client software for each game, where you have part of the game running on the local device and part running remotely.
Any security minded individual should be VERY concerned with the data they are gathering and sharing. I don't want this, I understand some people do, but I don't. I can safely say if this is the...
Any security minded individual should be VERY concerned with the data they are gathering and sharing. I don't want this, I understand some people do, but I don't. I can safely say if this is the future of gaming, then I will sadly have to walk away from that future. I'll be sticking with my gaming PC and low latency in-home streaming for the foreseeable future.
Edit: Also, this will be the beginning of "gaming/entertainment" specific internet service packages. Google knows this, Verizon knows this, cox, att, spectrum/time Warner, they all know this and I promise you Google was talking to them about deals before anyone else.
sounds great, and I love the idea of cloud-based technology, but my gaming library is on Steam. so it's not exactly like they're giving me access to my gaming library across a range of different...
instant access to your gaming library across a range of different devices
sounds great, and I love the idea of cloud-based technology, but my gaming library is on Steam. so it's not exactly like they're giving me access to my gaming library across a range of different devices... more like I'd have to buy a brand new one.
I don't know if consumers are actually going to buy into this product. So many companies have tried to do the whole streaming thing, but because of latency it's never going to be a thing.
I don't know if consumers are actually going to buy into this product.
So many companies have tried to do the whole streaming thing, but because of latency it's never going to be a thing.
A little tongue-in-cheek, but: I have this gut feeling that a product's name is often an indicator of how well it does. If it sounds stupid, it probably was rushed through some marketing phase too...
A little tongue-in-cheek, but: I have this gut feeling that a product's name is often an indicator of how well it does. If it sounds stupid, it probably was rushed through some marketing phase too quickly. Stadia sounds stupid. You're not "playing on Stadia", you'd never say that out loud. The "-ia" is some weird, latin appendix, "stad-" like in "stadium"? It's awkward. Same like "Ouya", "Zune", "Ngage" or "Pippin". It's a sign that no one ever took this seriously enough to care about the details.
Yes, "stadia" is the Latin plural of "stadium", the same way you go from "medium" to "media". I spent way too long debating over a name for Tildes and honestly, I think most names are terrible if...
Yes, "stadia" is the Latin plural of "stadium", the same way you go from "medium" to "media".
I spent way too long debating over a name for Tildes and honestly, I think most names are terrible if you focus on them in isolation, we just get accustomed to them. The most popular site in the world is named "Facebook", that's a pretty objectively terrible name (yes, I know what a face book is, but it's still not a good name). Google is an accidental misspelling of "googol", a word hardly anyone knows. Nintendo named a system "Wii" and then even named the next one "Wii U", double awful. You just get used to them all and stop noticing how bad they are.
They also posted a very long, detailed interview: https://www.eurogamer.net/articles/digitalfoundry-2019-google-stadia-phil-harrison-majd-bakar-interview
This can't be the only future of gaming, nor can it make expensive client hardware obsolete, because the entire competitive games industry will absolutely not be willing to add that much latency...
This can't be the only future of gaming, nor can it make expensive client hardware obsolete, because the entire competitive games industry will absolutely not be willing to add that much latency to their games, and VR is in the same boat.
And no matter how good their network gets, lowering latency is limited by the speed of light at an absolute minimum. The Earth is pretty big, and unless Stadia has servers running in every city, there will be unacceptable latency for fast-paced games and VR.
Hardware manufacturers market a 1ms difference in latency on their new monitor or a marginally faster polling rate on their mouse. I'm pretty sure adding tens of ms onto your game will always be a huge no-go for a large segment of the gaming industry.
Relevant table from article:
What is this latency based on? Is this a latency to get a frame on screen? Because these numbers don't make sense.
According to the table "Display" lag doesn't exist for the Stadia which makes no sense.
If you have a 60fps PC game, on a 6ms latency monitor you're going to have 1/60 = 17ms + 6ms = 23ms of latency.
So where's the 100-23 = 77ms of lag coming from?
Also what's going to happen with online games? So you have this 166ms latency which I assume the latency to get a frame on screen, then you'll have 100ms of network lag in a multiplayer game so you could end up with over 250ms of lag.
EDIT: I figured out they were using the latency of a PC over WiFi to another device, which I personally notice the latency.
Oh wow, that's kinda a weird comparison, lol.
That's very encouraging. Got my fingers crossed for the release later this year.
The best use case of this tech if we can massive reduce the lag would be for VR.
Unfortunately, I don't think that will ever be possible. VR is extremely latency-sensitive, and you will get sick if the view doesn't move correctly and immediately when you look around. I'm quite accustomed to VR at this point and regularly play games considered "high intensity" that make others feel motion-sick, but anything that lags/hitches even briefly while I'm turning my head will still make me start feeling sick almost instantly.
Years ago, Michael Abrash wrote that you needed a latency of 20 ms or even less for VR:
Since cloud gaming requires a round trip (you send your input to the server, it processes it and sends back the result), that means you actually only have half that long in each direction, and it's not possible to send data over the internet anywhere close to that quickly.
To add: The tested setup was 1080p30fps on a 200mbps connection. Extrapolate a little, and that's ridiculous amounts of bandwidth required for the refresh rates and resolutions that a good VR experience requires. Even if you completely discount the latency, just getting the bandwidth to do anything VR-like will be very difficult for most users.
That really depends on your connection and the location of the servers you're connecting to... my buddy in Virginia has FIOS and regularly gets pings of <20ms on the East Coast servers for Rocket League. We often joke that the servers are in his basement. But yeah, overall 20ms is unlikely to be achievable for the vast majority of people in most cases, at least for the foreseeable future.
That's also pure network ping too though, without the other pieces of latency (input, the game processing the input, the display processing the stream data, etc.) and at some point you even start running into the speed of light as a limit. For example, let's say that you're allowing 5ms of network latency in each direction, and let's give 1ms of that for time spent going through routers and such on the way. That leaves 4ms for traveling through the physical cables, and if you use 124,188 miles/sec as the speed of light in fiber, that means the server you're communicating with can not be more than 497 miles (800 km) of cable away from you.
For the other pieces, a game running at a constant 100 frames/sec is spending 10ms on each frame, but even that is probably too much if you need latency this low. So you need a server capable of running the game at 200 FPS or so within 500 miles of every player, and even that's already pushing 15ms without including multiple other sources of latency.
It might be theoretically possible, but very, very unlikely to ever happen in realistic conditions.
That's not to rule out streaming technologies completely. A dual systems where the scene (or parts of the scene) are rendered locally, but much of the heavy lifting ("game logic") is handled server-side is still possible. It might rule out complex lighting or other expensive operations, but I can see a future there.
Think a more advanced version of Oculus' asynchronous timewarp.
Yeah that is what I was thinking. Do light computing on a phone like device and then do ray tracing or something super compute heavy in the cloud and bake that in or something.
What about a sort of asynchronous spacewarp along with cloud rendering? Have a local computer render the old frame and movement data and have a cloud computer render the graphically intense frames?
Maybe, but that's getting pretty far away from a game-agnostic streaming system and into custom client software for each game, where you have part of the game running on the local device and part running remotely.
Any security minded individual should be VERY concerned with the data they are gathering and sharing. I don't want this, I understand some people do, but I don't. I can safely say if this is the future of gaming, then I will sadly have to walk away from that future. I'll be sticking with my gaming PC and low latency in-home streaming for the foreseeable future.
Edit: Also, this will be the beginning of "gaming/entertainment" specific internet service packages. Google knows this, Verizon knows this, cox, att, spectrum/time Warner, they all know this and I promise you Google was talking to them about deals before anyone else.
sounds great, and I love the idea of cloud-based technology, but my gaming library is on Steam. so it's not exactly like they're giving me access to my gaming library across a range of different devices... more like I'd have to buy a brand new one.
It's also the ultimate DRM. This is just "always online" in a more convenient package.
if they don't allow mods, it's even less consumer friendly. first they came for our dedicated servers. then they moved the whole game onto theirs.
I don't know if consumers are actually going to buy into this product.
So many companies have tried to do the whole streaming thing, but because of latency it's never going to be a thing.
A little tongue-in-cheek, but: I have this gut feeling that a product's name is often an indicator of how well it does. If it sounds stupid, it probably was rushed through some marketing phase too quickly. Stadia sounds stupid. You're not "playing on Stadia", you'd never say that out loud. The "-ia" is some weird, latin appendix, "stad-" like in "stadium"? It's awkward. Same like "Ouya", "Zune", "Ngage" or "Pippin". It's a sign that no one ever took this seriously enough to care about the details.
Yes, "stadia" is the Latin plural of "stadium", the same way you go from "medium" to "media".
I spent way too long debating over a name for Tildes and honestly, I think most names are terrible if you focus on them in isolation, we just get accustomed to them. The most popular site in the world is named "Facebook", that's a pretty objectively terrible name (yes, I know what a face book is, but it's still not a good name). Google is an accidental misspelling of "googol", a word hardly anyone knows. Nintendo named a system "Wii" and then even named the next one "Wii U", double awful. You just get used to them all and stop noticing how bad they are.
"Tildes" is perfect, no worries! :D
They also posted a very long, detailed interview: https://www.eurogamer.net/articles/digitalfoundry-2019-google-stadia-phil-harrison-majd-bakar-interview
This can't be the only future of gaming, nor can it make expensive client hardware obsolete, because the entire competitive games industry will absolutely not be willing to add that much latency to their games, and VR is in the same boat.
And no matter how good their network gets, lowering latency is limited by the speed of light at an absolute minimum. The Earth is pretty big, and unless Stadia has servers running in every city, there will be unacceptable latency for fast-paced games and VR.
Hardware manufacturers market a 1ms difference in latency on their new monitor or a marginally faster polling rate on their mouse. I'm pretty sure adding tens of ms onto your game will always be a huge no-go for a large segment of the gaming industry.