By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Microsoft - Cloud Processing won't mean a thing for the Xbone

 

The Cloud will...

Significantly improve games 63 12.57%
 
Improve some games somewhat 65 12.97%
 
Might be used well here and there 101 20.16%
 
Merely PR BS 271 54.09%
 
Total:500

To sum EG's article, from things that a typical game engine cycle consists of:

- Game physics (update models)
- Triangle setup and optimization
- Tessellation
- Texturing
- Shading
- Various render passes
- Lighting calculations
- Post effects
- Immediate AI
- Ambient (world) AI
- Immediate physics (shots, collisions)
- Ambient physics

things that seem suited for running on cloud:

- ambient background tasks
- some forms of lighting (though it seems only for static environments)
- background AI

So, according to EG, not that much really can be co-processed via cloud, though it obviously can benefit open-world games with background AI/physics.



Around the Network
Jazz2K said:
I don't know more than anyone else on this forum, yes I'm not sure what to think of this but I'm also curious about it. You guys can't explain how this works either, then how about we wait for them to explain at E3 (hoping they will)... this is a reason a lot of people will not bother trying to explain something they don't know themselves...


The skepticism surrounding Microsoft's claims of boosting the Xbox One's performance by 4x via their cloud solution is well founded, and laid out in fairly easy to understand terminology in the Eurogamer article. Give it a look.

- "If you have the heart of a true winner, you can always get more pissed off than some other asshole."

It is way too soon to tell.

Distributed computing has been done successfully in the past, it just can't be done with certain aspects of gaming. However it can be done.

To successfully take advantage of distributed computing, the console or the application/game would need to measure the latency. If the latency of the connection is within a given tolerance, then it could use it for those applications where it is suitable.

What this also does is put the onus on broadband providers to reduce latency and increase network efficiency. One immediate way, would be providing a back-end channel that allows a direct communication with the broadband provider and Microsoft's own data center. This would mean Microsoft uses multiple back-ends to the Internet, and those back-end providers offer a single hop to each of those data centers. To go to Microsoft.com, which requires going through AT&T to get to them, I hit a wall in Chicago with 200ms+ with my current connection, then bad hardware in Minneapolis, Minnesota where every attempt beyond that point is met with either high latency times or dropped connections. If connections are routed directly based on connection proximity to local data centers, which Microsoft has in Chicago, than that latency is significantly reduced. For instance, right now my latency would be under 30ms, within the time frame to do just-in-time calculations.

But that only solves one aspect of latency. The other, bigger problem is the number of hops from the consumer to Internet. For some connections, the number of hops and the amount of latency between those hops is ridiculous. The connection I am currently on is less than 25ms per hop to the service provider, however another connection I use through AT&T sees multiple hops with a latency of 600ms + That's almost a second of wait time just for one tiny packet. One bad router can slow the works down to a crawl, and often an broadband service provider/network backbone provider, won't change out equipment until a higher failure rate occurs across the device. Granted, they can shut-off ports, but for them a slow port is better than no port.

So, yes. The current usefulness of distributed computing is low, it isn't entirely useless. It can be used in those instances where tasks are low latency. However, network performance will also be extremely important. Knowing what your current latency is between you and your network provider, and you and the destination (a little harder to do with Microsoft) would go a long way to determining how useful it would be for a particular individual. Implying that it is useless for everyone, everywhere is not correct. The closer you are to a Microsoft data center, the fewer hops you have to get to the Internet from your home, and the fewer the hops your connection takes to get to that data center, the better your chances are of it working well for you.

As with the power of computing, I think as we progress into the future the power of distributed computing will increase, making it more viable for distributed computation. It will never replace the console itself. Just like it can't replace the PC. But it can provide additional computational power and I think in time it will.

The only question is Microsoft trying to do something well before it can be reasonably achieved in the real world? I think like a broadband only online service in 2002, it is a risk but not impossible. The majority of homes in 2002 used dial-up Internet. Roughly 10% at the time had broadband Internet, fewer still 1Mbps or higher download transfer speeds.

Worst comes to worse, Microsoft has 300,000 servers available for either gaming, it's cloud-based services (Azure) or it's more traditional Internet-based services (messaging/e-mail).

to go to Microsoft.com, which requires going through AT&T to get to them, I hit a wall in Chicago with 200ms+, then bad hardware in Minneapolis, Minnesota where every attempt beyond that point is met with either high latency times or dropped connections.



From what I understood from the Digital Foundry article, this mostly PR right now, with a lot of issues to overcome. May be used mostly for no lag intensive computations, so mostly pre-baked things, rather a step backwards from the current trend of doing as much real-time elements as possible.



So it is happening...PS4 preorder.

Greatness Awaits!

ViktorBKK said:

There is a lot of buzz lately about the how the cloud will change everything, not only in gaming but in general.

What many seem to ignore is that internet infrastructure does not improve at the same rate as silicon. The cost of GPU power for example, has massively improved in the last 5 years. The cost of bandwidth on the other hand, has improved only slightly, and in some areas of the world, not at all.
In the long term, it is a lot cheaper to buy the latest hardware and perform your computing tasks locally, than it is to use a proxy server. The operating costs of bandwidth used & server costs in a cloud computing scenario, simply outweigh the initial investment and electricity costs of hardware that is used locally. This is especially true for graphics rendering because massive amounts of data are involved, and thus, even more bandwidth is needed.

I expect the cloud to remain a data storage service in the future and nothing more. If/when massive breakthroughs occur in the development of internet infrastructure, then things could change.

Agreed. This is a problem and will be the biggest problem for the cloud in the short/midterm. Also latency. How is real time computing distributed between Console+Cloud even possible right now. The Cloud can not actually improve graphics, IQ or Physics. The Latency that will be introduced in the rendering process is astronomical. Its simpler to just do the work at one place and send the result via the Internet to the User (Onlive) then to split up very timecritical tasks (gaming) and use distributed computing with the Internet as bottleneck.

It could be good for Milo style Games though. Where the AI is incredibly sophisticated and runs on a Server but sees you through Kinect and interacts with you through the Xbox.

I still believe we wont see alot of meaningful uses if any for the next 2-3 years. Cloud is the future but its far away we need way better Internet for that to work on a broad scale. 

 

 





Around the Network
thranx said:
I would love to see it in action.

http://www.datacenterknowledge.com/archives/2013/05/24/microsoft-will-back-xbox-one-300000-servers/

"Booty says cloud assets will be used on “latency-insensitive computation” within games. “There are some things in a video game world that don’t necessarily need to be updated every frame or don’t change that much in reaction to what’s going on,” said Booty. “One example of that might be lighting,” he continued. “Let’s say you’re looking at a forest scene and you need to calculate the light coming through the trees, or you’re going through a battlefield and have very dense volumetric fog that’s hugging the terrain. Those things often involve some complicated up-front calculations when you enter that world, but they don’t necessarily have to be updated every frame. Those are perfect candidates for the console to offload that to the cloud—the cloud can do the heavy lifting, because you’ve got the ability to throw multiple devices at the problem in the cloud.” This has implications for how games for the new platform are designed."


Thanks for that. About god-damn time someone gave a sober pro-cloud-computing response. Even if you are not necessarily pro-c-c. 

Queries:

1.1 What type of performance disparity will there be between those that are able to access the cloud (i.e. online) and those that aren't? The game needs to be able to run completely offline, so that implies a hell of a lot of extra work for devs.

1.2 In the same vein, the bandwidth-scalability factor is surely also a massive one. What if the latency/bandwith is not low/big enough to handle the entire set of instructions that the devs would want to upload to the cloud? This would imply a very, very complex, un-console-like software development process which I simply do not see happening. EVER. (I hope whoever reads this gets what I mean?)

1.3 The above two points shout "IT'S NOT WORTH IT" directly to my face.

2. I don't see how this whole development isn't completely open to everyone else as well. Sure, 300 000 servers need a hell of a lot of money (which MS has and Sony perhaps not), but there is no part of this ordeal which is limited to MS. Sony has server farms too.

3. Even experts (and YES, Eurogamer are experts - face it) have severe doubts about this whole thing.

 

I could go on, but I suspect few will even read these points, so I'm just hoping someone nitpicks me a bit. 

SEE EVERYONE, this isn't a flame-bashing-troll-thread! 



COKTOE said:
Jazz2K said:
I don't know more than anyone else on this forum, yes I'm not sure what to think of this but I'm also curious about it. You guys can't explain how this works either, then how about we wait for them to explain at E3 (hoping they will)... this is a reason a lot of people will not bother trying to explain something they don't know themselves...


The skepticism surrounding Microsoft's claims of boosting the Xbox One's performance by 4x via their cloud solution is well founded, and laid out in fairly easy to understand terminology in the Eurogamer article. Give it a look.




The ones that take my curiosity are Avalanche Studios

http://www.videogamer.com/xbox360/just_cause_2/news/xbox_one_cloud_is_one_area_console_has_advantage_over_ps4_says_avalanche_tech_lead.html

The Eurogamer Article comes from ArsTechnica... which in no way says MS approach with their 4X claim is BS

One guy says exactly what I think of all of this...

"The way I understood it (or rather how I would design it), is that the xbox1 is capable of doing the calculations in real time all on its own, but that means that it is running at full CPU usage all the time when adding in all the other calculations being performed. This can work, but is inefficient, wastes power, causes excessive wear on the components, and generates excessive noise and heat. The ideal thing to do is take a portion of the calculations, run them on a remote server where they will finish far faster, and then load the results into memory, so that some of the load can be removed from the console CPU.

In order to figure out which calculations to off load we need to categorize the calculations. Obviously some calculations can never be offloaded because they are direct reactions to your controller input (running, shooting, driving, etc.), and then some calculations can be cached because they never need to be recalculated (terrain, menu animations, cut-scenes, etc.). Then there are calculations whose initial conditions are variable, but whose variability is fixed over a short time frame; stuff like room/terrain lighting, or weather conditions, or NPCs walking around a market. Calculating the room shadows, or how the sun hits a waterfall cannot be done ahead of time if a game has an open structure and a day/night cycle. However, once the initial conditions are set (time of day, wind, rain, etc.) they are going to change according to an algorithm (or several algorithms), and that means that they can be precalculated by a remote server.

So the best way to think about this remote processing thing is not that the graphics quality is going to jump up and down, but that the CPU/GPU fan will be able to slow down/turn off more often which will actually improve immersion.

Keep in mind that this is just how I would design it, but I would be surprised if the engineers at microsoft didn't come up with a similar approach."

 

This is what I understand of it all but will wait for MS to explain more in depth at E3 or after...



Dr.Grass said:
I was actually hoping people would come in here and challenge what I'm saying. Instead, all of you seem to have the same opinion.


MicroSoft just said that this is what Gaming will be like on XBone with cloud

 



 



maximrace said:
Lol @ OP. Why should we come in here and prove you of your wrong? You already have your opinion and you seem to think you can't be wrong while you're already proven to be wrong. Avalanche ( an actual dev WOW) said there are some things they can do with the cloud. It maybe isn't much but it is something. So I voted for the third option because I believe an actual dev above some random biased forumposter

 

Dunno what to say to you... Maybe I'll just spew it all out: Avalanche is shit, I'm aware that some minor things are doable on the cload and I'm certainly not averse to discussing them (see my post to thranx). Maybe we should stop talking about our opinions here and just put quotes from devs (who all seem to disagree on something). You've contributed nothing and ignored some of the legitimate points made both in this thread and by for instance Eurogamer. And, you've blissfully ignored the link to the pro-cloud-computing link IN THE FRIGGIN OP. Now how am I not an ass for not reporting you?



HoloDust said:
To sum EG's article, from things that a typical game engine cycle consists of:

- Game physics (update models)
- Triangle setup and optimization
- Tessellation
- Texturing
- Shading
- Various render passes
- Lighting calculations
- Post effects
- Immediate AI
- Ambient (world) AI
- Immediate physics (shots, collisions)
- Ambient physics

things that seem suited for running on cloud:

- ambient background tasks
- some forms of lighting (though it seems only for static environments)
- background AI

So, according to EG, not that much really can be co-processed via cloud, though it obviously can benefit open-world games with background AI/physics.


Awesome post. Thanks.