MS: Xbox One 40x More Powerful Than 360 with the Cloud, Only 10x Without the Cloud

Do you think they really plan to outsource parts of the rendering to the cloud? If they really are, no wonder you have to be connected 24 hours.
 
So are they saying that the XBone will work like a reverse Seti? it'll send data to the severs to be processed and sent back, instead of receiving data to be processed and send back? And it's going to improve performance of the console by a factor of 4?

Highly suspect claims.
 
Dude... Stop defending the bullshit.

I have fibre, it sends me data at 4.5MB/s maximum can you explain to me how the fucking hell MS is going to fit all that data in my pipe, which is a big pipe in the grand scheme of things.

Some of the most strenuous calculations end up with small results

For physics, imagine an explosion occurring beside a car. server can calculate how much explosive force shifts the vehicle while calculating gravity, the stresses on the vehicle suspension and how the car will end up. A lot of calculations. (I'm ignoring things like glass and deformity for this example)

In reality the important data which reflects what's on screen is the end result; the location of where everything is. So that's like 5 XYZ coordinates.
 
They will probably elaborate more on the Cloud at E3, but it does indeed sound very promising. Or if they can give dedicated servers to all of the games released for the console then that's good enough for me.
 
Some of the most strenuous calculations end up with small results

For physics, imagine an explosion occurring beside a car. server can calculate how much explosive force shifts the vehicle while calculating gravity, the stresses on the vehicle suspension and how the car will end up. A lot of calculations. (I'm ignoring things like glass and deformity for this example)

In reality the important data which reflects what's on screen is the end result; the location of where everything is. So that's like 5 XYZ coordinates.

And thats going to happen in the cloud is it? Divide 4.5MB by 30 and thats how much data MS will be able to send me per frame in a 30fps environment provided latency or packet loss isn't a thing.

Deluded.

Just so you know modern games are currently beginning to hit around 2GB of unique assets per frame and the cloud is going to send me 150KB of that.... FUCK YEAH!!!

Better hope no-one else in the house is using the net when I game for maximums Cloud POWAHS!!
 
The sad part is that people, here even, don't understand that a lot of online business applications and physics-intensive algorithms are run entirely in the cloud today.

It's not conceptual, it just hasn't been applied to games yet. It's not even left field thinking. It's just a logical progression.

Will we see it soon? Probably not. Will it take years to culminate into something that we realize has value like Xbox Live did so many years ago? Probably, and maybe it won't amount to anything. Does anyone here have a clue what they're talking about. Definitely no.


I guess that includes people who seem to have blind faith in MS , not only the skeptical about the cloud making Xbox One 3X more powerful by sending data via ADSL
 
They will probably elaborate more on the Cloud at E3, but it does indeed sound very promising. Or if they can give dedicated servers to all of the games released for the console then that's good enough for me.

The Cloud means shit all unless Microsoft is coming to install fiber internet in all our homes, with perfect up time. The only real benefit there is to the Cloud is save transferring, and we already have that.
 
And thats going to happen in the cloud is it? Divide 4.5MB by 30 and thats how much data MS will be able to send me per frame in a 30fps provided latency or packet loss isn't a thing.

Deluded.

I just gave an example, yes?

The example I gave would be like 2 KB/s, if that.

C'mon, at least try.
 
Regarding light maps...

some people here have repeatedly stated that it's too large to send in a time sensitive manner to players.

So... here's how I would do it if I was in charge of the architecture.

1. Compute high resolution light map.
2. Create multiple LOD versions of it.
3. Send user high resolution light map information for closest areas.
4. Send them lower resolution light maps for further out areas
5. Send them higher resolution light maps as bandwidth becomes available.
6. Keep light maps on server, and send those to XBones with same area/lighting condition requests.

So, it'd be handy for something like slow dynamic lighting - time of day and weather effects.


@kitch9: you have a fundamental misunderstanding of how things work if you believe it needs to send you information for every frame, much less send you information in real time.
 
If the cloud is so powerful. Why not just add it to 360 and call it a day.

They could, in all honesty. But it still doesn't help that the 360 is almost 8 years old and has a lot of limitations.

You are still bound by the rendering GPU, and you still need a bunch of CPU time to process the data coming in.
 
10m xbone's are never going to be online at once. some small fraction is.

unless you think everybody in the world who owns an x360 is playing it 24/7. great logic.

no, 95% of them are sitting under the tv turned off at any given time, just like your ps3 is right now while you're posting on gaf.

Oh and btw, didn't they say 40X more than 360, not 40X more than Xbone? Yeah, they actually did. It'
s 4X Xbone, so redoe your math with division by ten for starters. Good thing we can keep our facts so straight while bashing MS for dishonesty.

So how do you expect their big game launches to go after your main defense is that not a lot of people are on at the same time? How will it work when people are all putting in Halo 5 for the first time with this hypothetical cloud doing some of the work? Are they going to have extra servers just in case for those moments even though there have been server problems in the past with new games' launches?

And even still, how does this work in places where the internet speeds aren't even fast enough to do this on the fly with no latency issues? Even the average speed in America is pretty slow so how do you tell part of your consumer base that even though they could normally play games online, they have too bad of internet connections to use the features that they keep proclaiming drastically improves the console's power? You can't develop a game to depend on the cloud so much unless they're willing to give up on some of their loyal fans who can't get better speeds in their area?

There's way too many gray areas to say that this is going to somehow improve the console drastically because they're going to need some sort of baseline cut-off that's even feasible without losing even more of their target audience when they've already alienated a lot of people in the world. People say they focus on America, but if that's truly the case, then they can't be too dependent on the cloud until speeds drastically increase in the country enough to where there's a huge market for it.

OnLive was supposed to revolutionize gaming too and they came out way too early to get a hold of the market. People aren't going to want to sacrifice low latency just for a performance boost if it makes the controls unresponsive like OnLive did for a lot of people...

The CLOUD might sound like a savior, but they better know what the limits are based on realistic internet connections unless they plan to only please the Adam Orth's of the world. "Not fast enough internet for THE CLOUD? Deal with it and the disadvantages in games."

I'm not writing it off until E3, but I don't get how people aren't allowed to be skeptical of this. That doesn't make them an Xbox hater just because this sounds like PR talk or unfeasible for the international market. Skeptical =/= Hater, especially when you're making assumptions too just on the other side of the discussion. No need to get hostile just because some people rather not blindly accept bold claims, especially when they've made a lot of PR statements recently that contradict other statements or avoid the questions or problems at had entirely.
 
I wonder why it hasn't been done before if its this simple? Why do we need 8GB of 40TF of powah can be sent down 2KB?

YUp, i wonder why Nvidia released Titan GPU for example, or why Qualcomm is developing better processors if cloud is going to be the ultimate cpu/gpu. You know, just 2kb of data up&down and done, Crysis 3 running at 1080/60fps on any device
 
I wonder why it hasn't been done before if its this simple? Why do we need 8GB of 40TF of powah can be sent down 2KB?

It *has* been done before, but not at scale. To support enough concurrent users, you'd need a huge farm of servers and no publisher would be willing or able to take on that investment. But MS has Azure which is already ready to go to support such a thing at the proper scale. If they offer it to every game as a service, you can imagine it would at least be considered.
 
Regarding light maps...

some people here have repeatedly stated that it's too large to send in a time sensitive manner to players.

So... here's how I would do it if I was in charge of the architecture.

1. Compute high resolution light map.
2. Create multiple LOD versions of it.
3. Send user high resolution light map information for closest areas.
4. Send them lower resolution light maps for further out areas
5. Send them higher resolution light maps as bandwidth becomes available.
6. Keep light maps on server, and send those to XBones with same area/lighting condition requests.

So, it'd be handy for something like slow dynamic lighting - time of day and weather effects.


@kitch9: you have a fundamental misunderstanding of how things work if you believe it needs to send you information for every frame, much less send you information in real time.

So how many TF would that save?
 
Are gamers really as gullible as MS believes they are? You people sure are lapping up anything to defend inferior specs... Pipe dream, especially for anyone outside the USA. Pipe dream there too, but at least broadband is more common.
 
You do know how the cloud works right?

By NPD there are probably what, 30m 360's sold in the USA? How many actually are turned on and playing right now? 2m?

What about at 1AM in the afternoon on a school day?

That means you can be a whole lot more judicious with your computing power with cloud. If you dedicate 4X for all those 360's in the above example, that's still 8m rather than 30m.

Plus it's probably a whole lot more efficient.

This whole thing is pretty confusing. Gonna be fun though.

The damage control in here is hilarious. Everybody so skeert.

Holy shit. 1am in the AFTERNOON.
 
Dude... Stop defending the bullshit.

I have fibre, it sends me data at 4.5MB/s maximum can you explain to me how the fucking hell MS is going to fit all that data in my pipe, which is a big pipe in the grand scheme of things.

If anything it should take much less data than sending a frickin' video stream like Online and Gaikai do. They don't have to send you massive amounts of data, all the game data is obviously stored in the cloud directly on their servers. Direct video is about the dumbest and most bandwidth intensive solution for cloud gaming, it's the brute force solution. I mean ... jesus. If you have a fancy game with realtime weather and wind system that would normally be calculated on your end. With the cloud all it takes is to send one KB of data per second to tell your console what the actual conditions are where you currently are. They could even take all the players around the world, add them to the equation and send the results to everyone. Like a butterfly effect.

AI can be done by the cloud, it's already done so many times. Left 4 Dead, who do you think calculates those Zombies when you're joining a server? It's not happening on your end, you just receive their positions. Every P2P connection over Live is basically the same thing, the host calculates, everybody else just receives. The cloud concept doesn't change what is already there, it just takes it to dedicated servers with a much better bandwidth. If you can do P2P gaming with as low as 15 KB/s I'm pretty sure cloud gaming with 4.5 MB/s should not be a problem.

There is stuff where it's not viable to use the cloud, time critical stuff. If you make the cloud render dynamic lighting it's gonna suck, who wants to have the shadows and lights around you adjust 200 ms after you turned? But it is possible for global illumination and other things where you aren't in direct control and thus have no relation to when stuff should happen on your screen.

Azure is also a much cheaper solution than game streaming. With streaming you need to basically have one server for every concurrent player. Something like a fully dynamic weather systems needs to be only calculated by a couple of servers and then sent to all the clients. It doesn't take as much computing power and it doesn't take as much bandwidth.

Or pedestrians in a game like GTA. You have to calculate every single one of them. Which means they basically only exist while you see them/are in a certain area around you. Ever wondered why people or cars disappear? With the cloud you could actually calculate all the positions of everyone in a city of 50.000. It's a wholly new game for crowd mechanics. All you would ever get are the positions of everyone you see, where today you have maybe 30 active people around you it could be 2.000. Rendering would not be a problem thanks to modern techniques like tesselation.

So yeah, cloud gaming is practically limitless for certain tasks. Anyone who denies that the cloud can have a severe positive impact on video games is kidding themselfs. And compute hours are cheap enough today to make it actually a pretty smart decision not to spend all that money on console power but to outsource to the cloud.
 
They could, in all honesty. But it still doesn't help that the 360 is almost 8 years old and has a lot of limitations.

You are still bound by the rendering GPU, and you still need a bunch of CPU time to process the data coming in.

You honestly have no fucking clue what you're talking about.

You should honestly be banned for spouting such nonsense as if it's fact.
 
I can't wait until people find out about the peer-to-peer processor sharing running off MiiBay that Reggie is going to announce at E3.
 
HA HA HA HA... man I had a good belly laugh there. Marketing folks everywhere jizz their pants when saying "the power of the cloud" nowadays. It's called "the power of recurring revenue" guys. Peddling this 40x bullcrap, get the f out of here.

PS: For those needing more proof: Just paging out of RAM into disk makes everything go to shit performance wise. Compared to that, a lookup on even the lowest latency Akamai peer is going to look like an ice age.

Educate thyself: http://www.eecs.berkeley.edu/~rcs/research/interactive_latency.html
 
Haha. They will keep talking about this and it will never become reality.
You can't rely on cloud computations for a realtime graphics pipeline. If so, why even release a new console...
Watch all multiplatform games be inferior to PS4 outings.

Because the PS3 is more powerful than the Xbox360 all multiplatform games look better in PS3 ... but in reality most multi-games look better on XBox360 and only some exclusives show the power of the PS3.
 
Regarding light maps...

some people here have repeatedly stated that it's too large to send in a time sensitive manner to players.

So... here's how I would do it if I was in charge of the architecture.

1. Compute high resolution light map.
2. Create multiple LOD versions of it.
3. Send user high resolution light map information for closest areas.
4. Send them lower resolution light maps for further out areas
5. Send them higher resolution light maps as bandwidth becomes available.
6. Keep light maps on server, and send those to XBones with same area/lighting condition requests.

So, it'd be handy for something like slow dynamic lighting - time of day and weather effects.


@kitch9: you have a fundamental misunderstanding of how things work if you believe it needs to send you information for every frame, much less send you information in real time.


Might as well compute those lightmaps BEFORE the game ships!

For a slow time of day effect you would need what, a couple dozen sets of lightmaps. This is NOTHING for a bluray disc! And will not require always online or a monthy fee to play a GD SP game
 
If anything it should take much less data than sending a frickin' video stream like Online and Gaikai do. They don't have to send you massive amounts of data, all the game data is obviously stored in the cloud directly on their servers. Direct video is about the dumbest and most bandwidth intensive solution for cloud gaming, it's the brute force solution. I mean ... jesus. If you have a fancy game with realtime weather and wind system that would normally be calculated on your end. With the cloud all it takes is to send one KB of data per second to tell your console what the actual conditions are where you currently are. They could even take all the players around the world, add them to the equation and send the results to everyone. Like a butterfly effect.

AI can be done by the cloud, it's already done so many times. Left 4 Dead, who do you think calculates those Zombies when you're joining a server? It's not happening on your end, you just receive their positions. Every P2P connection over Live is basically the same thing, the host calculates, everybody else just receives. The cloud concept doesn't change what is already there, it just takes it to dedicated servers with a much better bandwidth. If you can do P2P gaming with as low as 15 KB/s I'm pretty sure cloud gaming with 4.5 MB/s should not be a problem.

There is stuff where it's not viable to use the cloud, time critical stuff. If you make the cloud render dynamic lighting it's gonna suck, who wants to have the shadows and lights around you adjust 200 ms after you turned? But it is possible for global illumination and other things where you aren't in direct control and thus have no relation to when stuff should happen on your screen.

Azure is also a much cheaper solution than game streaming. With streaming you need to basically have one server for every concurrent player. Something like a fully dynamic weather systems needs to be only calculated by a couple of servers and then sent to all the clients. It doesn't take as much computing power and it doesn't take as much bandwidth.

Or pedestrians in a game like GTA. You have to calculate every single one of them. Which means they basically only exist while you see them/are in a certain area around you. Ever wondered why people or cars disappear? With the cloud you could actually calculate all the positions of everyone in a city of 50.000. It's a wholly new game for crowd mechanics. All you would ever get are the positions of everyone you see, where today you have maybe 30 active people around you it could be 2.000. Rendering would not be a problem thanks to modern techniques like tesselation.

So yeah, cloud gaming is practically limitless for certain tasks. Anyone who denies that the cloud can have a severe positive impact on video games is kidding themselfs. And compute hours are cheap enough today to make it actually a pretty smart decision not to spend all that money on console power but to outsource to the cloud.

So why is that unique to MS when others have done it before?
 
this might work....


at MS campus where the network is at least gigabit, the demand on the "cloud" is known and static, the network has low latency due to locality of servers and the servers probably are in the same room or building as the prototype xbox ones.

out in the wild?

no fucking way


and lets not forget

when the servers die, the games die
 
Are gamers really as gullible as MS believes they are? You people sure are lapping up anything to defend inferior specs... Pipe dream, especially for anyone outside the USA. Pipe dream there too, but at least broadband is more common.

Azure has been working fine outside of the USA for a while now. I don't know why computing power has anything to do with the amount of data transferred or how much bandwidth is required. It might take 100s of CPUs a few seconds to compute the best chess move, but the data transferred to the console would only need to be what the move was.

This is not like OnLive where every frame of the game needs to be sent down 30/60 times a second.

I think the main problem with this stuff is the lack of good examples of how this tech would be used in modern games. But this tech has existed for many years in non-gaming applications. So it's not a pipe dream there.
 
You honestly have no fucking clue what you're talking about.

You should honestly be banned for spouting such nonsense as if it's fact.

Aww don't be upset. Because you'll feel really bad when this does (and it will) come into fruition.

No need to be so blinded with your allegiances.

(And yes, I do know what I am talking about)
 
cloudpower.gif
 
Aww don't be upset. Because you'll feel really bad when this does (and it will) come into fruition.

No need to be so blinded with your allegiances.

(And yes, I do know what I am talking about)

You haven't provided a single source or explanation for ANYTHING you've posted.


Also, you better believe I'm going to put you on blast when you're wrong.
 
this might work....


at MS campus where the network is at least gigabit, the demand on the "cloud" is known and static, the network has low latency due to locality of servers and the servers probably are in the same room or building as the prototype xbox ones.

out in the wild?

no fucking way


and lets not forget

when the servers die, the games die

You are just jelous. aegis told me the secret jizz sauce works.
 
So why is that unique to MS when others have done it before?

Because MS has the scale and infrastructure already built. Many games have done this already, but it's expensive for them since they have to pay for everything. And it's a big investment. MS is providing all of that because they already have it built in Azure which is being used for non-gaming apps and has been since 2010 or so.

It might be worth reading about Azure and just substitute the word game anywhere you see the word app :)
 
AHAHAHAHAHAHAHAHAHAHAHAHAHA..... Ha!

That's a good one. What else? It'll run 8k resolution while outperforming the latest high end PCs for 20 years to come. Oh, yes. I'm sure.
 
It *has* been done before, but not at scale. To support enough concurrent users, you'd need a huge farm of servers and no publisher would be willing or able to take on that investment. But MS has Azure which is already ready to go to support such a thing at the proper scale. If they offer it to every game as a service, you can imagine it would at least be considered.

why would publishers want to pay for this? how does it make them more money?
 
If I ignore, for a moment, that the MS' handling of the Xbox One reveal has been ridiculously stupid, I don't think the cloud has ever been pushed so strongly and so without context.

Compare the usual cloud-supported device disclaimer of saying something like "you need Internet to use this feature" and what the Xbox Australia guy said.
 
Ok, I slept over that idea of cloud processing, and I think many people are overreacting, unsurprisingly. There is certainly something there, only you have to read between the lines.

First, I would dismiss all the straight "this is bullshit", "it cannot work", "pie in the sky" comments. There are a few things pointing to the fact that cloud processing will be a thing : Rare recruiting people working in that field for example, and MS claiming an upgrade from 5000 to 300000 servers (whatever that means, there is no point in such an upgrade for standard XBL services). MS has been talking about that cloud from the beginning, in keynotes etc, they wouldn't bother doing it if there wasn't at least a few things running on it. The question is what and how important it is.

Second, the "40x more powerful" is obviously PR talk (any quantitative measurement of "console power" is mostly subjective), but it also needs to be interpreted. Having a "console power" boosted 2x by remote processing doesn't mean your game will be twice as fast, or twice prettier, or whatever. Let's say that task A would usually require 10 magiflops of a console out of a total of 100, but you can send that task to a big server which can run a more advanced algorithm that uses 110 magiflops, then some will say you doubled the power of your console, even if it is only for a minor task. But 90% of the game is still unchanged and running on local hardware.

Third, what can it be used for ? Well, I don't know. But if I had to speculate, and from what people said in the technical keynote, it wouldn't necessarily be "traditional" tasks. Current games are designed for local processing, so most of the tasks running in them are not meant to be computed remotely. Some of them could work with a little latency, but I suppose they're not the biggest part of the whole program.
I think what cloud processing could bring to games won't be taking care of common tasks in current games, but allowing for new tasks and new features. Of course we'll need to wait and see if it really makes games better, or if it is just a nice option that will be used in a handful of first party games.
 
You haven't provided a single source or explanation for ANYTHING you've posted.

You don't need a source for this, one.

Secondly, if you're having difficulty understanding these concepts and my posts, swing by your local college and apply for a course in computer science or something
 
Top Bottom