-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
Digital Foundry: Doom Eternal on Stadia Looks Great - But The Lag is Just Too High
- Thread starter Beer Monkey
- Start date
I don't know how you'd interpret "delay on Stadia" to mean anything other than the delay, on Stadia, over and above Xbox One, especially when those words are being said while showing a video of Stadia response time being higher than Xbox One.
Even if you did misinterpret that and figured that DF were talking about end--to-end latency on Stadia being 79ms, if you knew anything whatsoever about latency, that shouldn't have made sense to you. 79ms of latency on Stadia would be an achievement that DF would be heralding as a giant positive, not pointing out as a negative.
Direct quote from the video:
"Using a 240 fps camera we get between 19 and 24 frame delay on Stadia in every test, from the trigger to the mussle flash on-screen. Whitin a 240 fps second, that means latency on Stadia comes in within the 79 to 100 ms range, averaging in the middle at 91 ms according to these tests."
They are literally saying in the video that the Stadia has 79-100 ms total delay, from button press to action. DF has fucked up here, since their text says something completely different:
" To cut to the chase, using a 240fps high speed camera, the difference between pressing the fire button and the action playing out on-screen sees Stadia adding an extra 79ms to 100ms over the same motion executed on Xbox One X. Yes, to be clear here, 79ms to 100ms is the extra latency you get by playing Doom Eternal on Stadia. Interestingly, the Xbox One X result is quite laggy in and of itself, with 94ms the most frequent result, but adding anything up to an extra 100ms on top of that and then adding display latency means that Doom Eternal on Stadia delivers cumulative lag over one fifth of a second. "
For context, a 300Mb/s connection in the UK means it's a Virgin Media connection, and they don't have a fantastic reputation for low latency connections. With that said, the extra 100-ish milliseconds of latency compared to the Xbox Series X version sounds roughly in line with other latency comparisons.
Is it an "extra" lag of 90ms on average, or is the whole lag 90ms in itself. Because if it's the latter than that's pretty darn good for a streaming game at 60FPS. Last gen we struggled to get 80-90ms even on 60FPS games.
However if it's an extra lag of 90ms on top of the base input latency, then that's an awful lot for a 60FPS streaming game, considering I expected stadia to add around 40ms of lag give or take on average as seen in some other games (server location dependent).
However if it's an extra lag of 90ms on top of the base input latency, then that's an awful lot for a 60FPS streaming game, considering I expected stadia to add around 40ms of lag give or take on average as seen in some other games (server location dependent).
This very well be the case but look at the chart over here (https://www.google.com/amp/s/www.eu...dry-2019-hands-on-with-google-stream-gdc-2019) It just says X Box one X, without any fps information.
It's just formatted very badly and filled with ** to somehow make it look like stadia = X box one X regarding lag.
Cant hyperlink, am on phone.
Last edited:
There is it's called, fq-codel or cake being used as your queueing mechanism, df nor stadia use it on their links. It's performance is peer reviewed. Both are older than the udp protocol used for stadia.
What's also been peer reviewed in the link is that us adding bandwidth in the amounts we have since the late 90's has actually made the internet slower. So bragging about bandwidth to solve latency doesn't show you wish to address lag or latency issues. They should be seen as seperate but again someone here has conflated something that has a distinct meaning. Lag is a variance of your latency. Latency is and will always be the speed information from machine to machine.
Overall I didn't learn much except the latency is mediocre, it's under 125 which is where real time responsiveness becomess a lot less than it typically is.
Stadia implementation of a network is a joke but the potential to do a lot with cloud will always exist if the one operating the network takes steps to address latency. Google has done some but not enough and not where it matters.
Thx for the info Delusibeta. Testing standards for stadia is joke as much as the network itself in some areas.
Because that's commonly referred to as "Putting words in someones mouth", something generally seen as impolite.
Look, I can see the video too, but what he said mixed up total-frame-delay and compared-to-X1X-delay with the 2 number sets. You could argue I should do the maths myself to figure it out, but come on.
And at the end of all this, neither the article nor the video bothers to give the total delay (in milliseconds) of the X1X and Stadia.
Would have been nice to have those numbers side-by-side in a table, like they've done before.
Y'know, "Just to be clear".
It would make some combat sections straight up impossible.
Just thinking of the final level with any kind of lag makes me shudder
Honestly, picture quality wise it looks pretty damn fantastic for a stream. Shame about the latency.
Yeah, the additional footnotes here are a mess, and were attempts by DF to make it clear that they were doing their best to try to compare things here, even though they weren't in a good position to do that. It still ends up showing additional latency on Stadia over the Xbox One X, without display lag factored in. They go on to say that Odyssey is not a great example for testing, and somehow in that test the Xbox One X has weirdly got much higher latency than the PC version running at 30FPS, so it could be that the near-equivalence in that test is partly due to some weird issue on Xbox One X.
They probably shouldn't have tried this kind of objective measurement since Stadia was pretty far from release at that stage and they had to throw in a bunch of caveats, but people were curious about the latency so I guess that was them trying to satisfy that curiosity any way that they could.
I don't think it's unreasonable to expect that anyone trying to make the kind of argument you were (that is, trying to correct someone who was posting about the video) would have at least a basic comprehension of the subject matter they were talking about. If someone doesn't know enough about latency to know that 79ms end-to-end latency would be a stunningly low number for a streaming service, then it's fair enough that they'd be confused and I wouldn't expect them to do the mathematics. However, for anyone trying to correct other people I'd normally expect at least basic familiarity with typical latencies of streaming services.
The article says "Interestingly, the Xbox One X result is quite laggy in and of itself, with 94ms the most frequent result". That's the total (typical) delay for Xbox One X, and since you know the average additional delay for Stadia, that gives you the baseline for the range of Stadia results as well.
I'd be really interested in seeing how the accelerating pace of residential full-fibre rollout impacts things going forward. A mate of mine moved into a new-build with full-fibre via Openreach and the difference is night and day, even though speeds are limited to 330 Mbps at the moment.
That's about as far as FTTP is at the moment, new builds where the developers have sold the infrastructure contracts.
They do. Here's a direct quote from the article:
Digital Foundry said:
This remains one of the funniest shit ever said by a gaming company
Stadia threads really are a shit show.
I get Google failed to deliver a compelling product but this is nothing but a step in better cloud gaming solutions.
Seems like people just want to be right about cloud gaming never being a viable option.
Hope that changes and this tech gets better.
I get Google failed to deliver a compelling product but this is nothing but a step in better cloud gaming solutions.
Seems like people just want to be right about cloud gaming never being a viable option.
Hope that changes and this tech gets better.
They will disappear once a credible example shows up and just remind people of the past.
Google did bite off a lot more than it could chew and negative latency is really dumb comment to ever make.
Having played Doom Eternal about halfway through, I have no idea how you could consider buying this if you thought there'd be a second of lag. It's a game about fast movement, twitch reactions and constant action. There is no way it would ever work with lag.
When I saw this video it reminded me that whenever I used the Stadia speed checker tool before, I'd get like 35 Megabits/second.
This always bothered me, because according to Speedtest.net, I'd get low 900s (sometimes high 800s).
I checked again after watching today's video and got
man, that would suck.
I guess this is COVID-19 related though, since Speedtest isn't giving me but like 60ish download (945 upload) right now too.
This always bothered me, because according to Speedtest.net, I'd get low 900s (sometimes high 800s).
I checked again after watching today's video and got
man, that would suck.
I guess this is COVID-19 related though, since Speedtest isn't giving me but like 60ish download (945 upload) right now too.
Makes me wonder just how promising cloud gaming can be for super fast response limited games. Racing games as well simply do not translate to cloud gaming (if you're playing remotely seriously).
It's fine for certain games but not suitable for everything. As a back up option to maybe play your home games while you're on your break at work it's fine, but I can't see it being my main way to play the big games I'm really invested in. I'd never play TLOU2 for the first time with cloud gaming for example, I'd make sure I was playing natively.
It's fine for certain games but not suitable for everything. As a back up option to maybe play your home games while you're on your break at work it's fine, but I can't see it being my main way to play the big games I'm really invested in. I'd never play TLOU2 for the first time with cloud gaming for example, I'd make sure I was playing natively.
Well, there goes my way of playing Doom Eternal any time soon since I feel like playing it on consoles isnt the right thing to do..
OP
OP
I live in a brownstone-ish sort of four-unit building built in 1890 and we got FTTP a year ago. Everybody gets their own ONT in the basement and their own fiber pull. Landlord was smart enough to pull CAT5e to units 12 years ago so it's easy to install.
However, this is a local ISP. I've got >800mbit down (actual), >200mbit up, no caps. I know my situation is atypical.
Trying to imagine that lag with a *mouse*. Barf.
That quote does not contain what I was talking about.
Read my post again.
I watched the video, read the article and read all your posts.
I'm good, thanks.
A second of lag would make any game completely unplayable. Thankfully that's not the case here, even with the decent amount of latency.
wow, you are really lucky. My house is a similar age on a street of similar properties and all we have is Virgin (albeit the 350mbit service!) which isn't quite as good for gaming
Yaaaay another thread to shit on Stadia, way to go guys.
Does anyone know if this DF guy (Thomas Morgan) is on Era? I'd like know how much he fiddled with the controls. Game was absolute shit to play on Stadia with the default control settings, main offender being the Look Smoothing option. I experienced the same Thomas was describing in his review that it was more efficient to get enemies in the crosshair by strafing rather than "panning the camera". Turning Look Smoothing all the way down to zero and lowering the X, Y sensitivity was a night and day difference.
Does anyone know if this DF guy (Thomas Morgan) is on Era? I'd like know how much he fiddled with the controls. Game was absolute shit to play on Stadia with the default control settings, main offender being the Look Smoothing option. I experienced the same Thomas was describing in his review that it was more efficient to get enemies in the crosshair by strafing rather than "panning the camera". Turning Look Smoothing all the way down to zero and lowering the X, Y sensitivity was a night and day difference.
The video suggested you can get used to the added lag. KInda like people learning fighting games online, you can adapt. It may not be the best way to play but its certainly not terrible.
If you actually want to to participate in a grown-up discussion here on Resetera, you need to change your attitude.
The part of of the article I quoted contains the total delay of both the Stadia and X1X versions of Doom Eternal. I will post the quote again and highlight the relevant sentences.
Digital Foundry said:
"over one fifth of a second" is 200ms+ of latency.
Well It took being an asshole to get you to reply properly.
I kid, I kid!
But sure, thanks for clarifying. I guess I need to go stick my head in a bucket of iced water and cool off, because clearly my reading comprehension has fallen off a cliff in the last few hours. Stadia threads are always such a shitshow they seem to bring out the worst in me.
Honestly, I shouldn't care about this stuff. My experience of playing Doom Eternal is what it is, and a DF video doesn't change that at all.
Man I wish they would hurry up and release the Base tier, so that people can have their own experience with it, instead of just these videos and hot-takes.
I don't dispute DF's numbers, but until people actually use it, that's all they are, some numbers. I doubt most people really know what 200ms really means.
In case you missed it, I apologised to you.
We cool?
Unless they start acting like riot they will never get to 10ms-35ms sweetspot for users if they did shit right. They don't build their infrastutcure to hold up during a load either. Riot got their act together but so almost nobody else runs their networks the same way in gaming right now which is pitiful. Until consumers with enough pressure force things nothing will happen.
Cloud doesn't limit networking, bad networking limits cloud or gaming.
laws of physics
laws of physics
Ye cannae change the laws of physics
laws of physics Jim!
I don't think xCloud or whatever Sony's equivalent (not counting PSNow because thats a game sub like Gamepass) will be doomed or judged as harshly, simply because those are/will be supplements built around existing and strong ecosystems rather than a new service not built around any ecosystem that is trying to replace traditional console gaming.
Stadia is "doomed" because it is trying to compete with the consoles but without having its own physical box
I would guess most Stadia users playing in 4k would be using a Chromecast hooked up via wired ethernet, because it requires (and ships with) a Chromecast Ultra, which has an Ethernet port built-into the AC adapter. I mean, why wouldn't they? Why don't you, when you have your consoles on wired Ethernet? One would think you'd be *more* likely to use ethernet on a streaming service than on a physical game console.
300mpbs and ethernet isn't "god tier" these days, I'd say it's "pretty good", at least in cities in the US (Comcast offers gigabit "nationwide" these days at like $95/month with cable TV, for example).
Can't speak to others but for me the router or an ethernet outlet isn't near the TV so I've been using it over wifi. Just did a speed test that puts it at 67.4Mbps down and 18.8 up and I've not had any issues with playing games on Stadia. Only game that did mess up was Just Dance but it was hard to know if it was the mandatory phone input fucking up or the service. Yet to try Doom Eternal so it could be an exception but Destiny 2 and others have been fine.
Just wanna make sure no wires are getting crossed here (because lord knows I've done it enough already!).
Its 6-9 frames from a 60fps game display. (The 240fps video just makes it easy to see). So if I'm getting this right, it should be:
6 out of 60 frames = 10% of a second, which is 100ms.
Consistent with DF's numbers.
So Stadia's total measured lag for DF is 20% of one second, or 200ms
Side-note: I'd like to strangle the person who decided to divide a second up into 1000 and call it a millisecond. You'd think it should be one millionth of a second, but nooooo, someone had to be difficult!
