• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Crackdown 3 uses cloud-based physics processing to enable environmental destruction

How does the statement that a PS4 is never going to outmatch 20x the Xbox One hardware by using GPGPU not make sense?

Physics are handled heavily be CPUs too and no a few GPU cores running GPGPU will never outmatch 20x CPUs.

Nobody said that. I said this doesn't make sense:


If part of a GPU could outmatch 20 CPUs we would be using CPUs in our computers.

If GPUs could offer more power why would we be using CPUs? I then said at given tasks a GPU can greatly outperform a CPU. Anyway I think I was pretty clear.
 
Nobody said that. I said this doesn't make sense:

If GPUs could offer more power why would we be using CPUs? I then said at given tasks a GPU can greatly outperform a CPU. Anyway I think I was pretty clear.

Maybe if you looked back a reply or two you'd see what I responding to.

You wouldn't think the statement that GPGPU can't outmatch 20x CPUs wouldn't start a argument.......

Great analysis.

Of what?

Common sense?
 
FordGTGuy said:
This is complete non-sense, the only reason Microsoft can do it Is because they are the second largest cloud company in the world.
You should tell that to all the startups that have been building similar systems for the past few years.

But anyway - for people championing this - look forward to a future more full of games-as-service models then ever - more entrenchment in online-processing will only make Gaf's "favourite" business models more dominant.
 
You should tell that to all the startups that have been building similar systems for the past few years.

Please, post a startup that is running cloud computations for a AAA title in North America, Europe, Australia and Asia with 20x the local processing power of a console.
 
Played this at Gamescom after a presentation by the team, I was seriously in awe by the physics involved. We took down a tall building in order to take down the bridge between two other structures next to it, not a single dropped frame.

I'v been very skeptical about cloud computing, but I can't argue that what I saw and played was very impressive.
 
Maybe if you looked back a reply or two you'd see what I responding to.

You wouldn't think the statement that GPGPU can't outmatch 20x CPUs wouldn't start a argument.......

You're speaking in riddles. What about the post you replied to? namely this:

The first part of that is obviously ridiculous. As for the second part I have no idea if a PS4 could do this, as I'm not a developer, but I seriously doubt it.

If you don't understand my comments don't try and deflect it and say I wasn't following the conversation.
 
Huh? What does that have to do with anything we're talking about? :P
It's like Folding@Home but in reverse. I forgot of name of the university but when people launched the application, the university's server siphoned a small percentage of the ps3's CPU allocation. With hundreds of thousands of people providing additional processing power, the university was able to run real-time simulations.

University server = Your Xbox One

Multiple ps3s = VMs on the rented server

Very easy analogy. Also it was relevant to your argument about other examples of cloud computing in other industries, this one being in the medical field. Folding@Home and cloud computing were a really big thing in the early ps3 days.


Nicole Gaynor said:
Pande's lab uses cloud computing through Folding@home and Google Exacycle to run many detailed molecular dynamics (MD) simulations of protein folding independent of one another.
http://www.ncsa.illinois.edu/news/story/cloud_super_computing_results
 
Played this at Gamescom after a presentation by the team, I was seriously in awe by the physics involved. We took down a tall building in order to take down the bridge between two other structures next to it, not a single dropped frame.

I'v been very skeptical about cloud computing, but I can't argue that what I saw and played was very impressive.

How much did they pay you to post this?

:)
 
It's like Folding@Home but in reverse. I forgot of name of the university but when people launched the application, the university's server siphoned a small percentage of the ps3's CPU allocation. With hundreds of thousands of people providing additional processing power, the university was able to run real-time simulations.

University server = Your Xbox One

Multiple ps3s = VMs on the rented server

Very easy analogy. Also it was relevant to your argument about other examples of cloud computing in other industries, this one being in the medical field.

Not really, I heavily doubt that Folding@Home is latency sensitive in anyway.
 
Its a mess in here.


Im all for this, Especially in a game like Crackdown but I always have to think that as more games get released that may partially require services in order for certain portions to work properly, what will that do for game conservation?

We know there are dozens of games that need online services and they can't run forever... What happens when that time is done? Nobody can play that mode, maybe even the actual full game again?
 
Crackdown's destruction is insane. I don't know how people can compare it to other games, there's no game that compares to the level of destruction shown in the video.

Exciting times ahead and I'm looking forward to see how much they push the cloud in games. I remember Kampfheld posting grass spline gif's, so it's not just destruction it can be used for. I hope to see more games take advantage of it in different ways especially if the outcome is anything close to as big as the CD destruction.

Speaking of Kampfheld will he ever be unbanned? Would be interesting to hear him out now the CD demo is released and everything he said about CD lined up so maybe he was working on it after all.
 
Please, post a startup that is running cloud computations for a AAA title in North America, Europe, Australia and Asia with 20x the local processing power of a console.

Are you kidding? One AWS c4.2xlarge instance is at least 20x the processing power of this generations consoles. Many, many, many startups are using AWS.
 
Played this at Gamescom after a presentation by the team, I was seriously in awe by the physics involved. We took down a tall building in order to take down the bridge between two other structures next to it, not a single dropped frame.

I'v been very skeptical about cloud computing, but I can't argue that what I saw and played was very impressive.

Thanks for posting your impressions. It sounds really impressive, can't wait to try it out in a Beta.
 
Saw the demo, the potential here is quite impressive. Not very familiar with the technology behind it, so looking forward to hearing more.
 
Are you kidding? One AWS c4.2xlarge instance is at least 20x the processing power of this generations consoles. Many, many, many startups are using AWS.

They are currently using this in a AAA title or in the next year globally?

You guys seem to be confusing whether the technology exists for it and whether or not these companies can afford to utilize it.(If the infrastructure exists in a way to offer the services globally too.)

It all depends on applications and costs and Microsoft is in a position to offer it while most couldn't or wouldn't risk it.
 
This is the key. How many developers will bother taking the time to make games with something that requires utilizing an always-online connection? That alone significantly shrinks your potential audience size. It's like the MotionPlus with the Wii, when given the time, we got some legitimately cool games that took advantage of it. But I can think of only two like that off-hand.

Yeah, because multiplayer games are really struggling with player population these days.

/s
 
FordGTGuy said:
Please, post a startup that is running cloud computations for a AAA title in North America, Europe, Australia and Asia with 20x the local processing power of a console.
Not sure where there was any mention of games that do this ("AAA" or otherwise), obviously nothing is on the market yet. And if Crackdown is first (which is likely) we're still 1-2 years out from seeing it in action over "NA/Europe & Asia".

Hockeymac18 said:
I think the scale of this software accomplishment is being overlooked/trivialized.
Gaf always trivializes any engineering accomplishments, but that's besides the point. If you ever looked at online development and cost-margins there, the notion that "cloud power" is free in any sense of the word is more than a little absurd (and I'm purely talking compute $ here, nothing related to complexity or ease of utilizing it).
The idea of running large scale game simulations entirely server-side is nice, but it pretty much spells the end of "pay for product" business model IMO (then again the writting has been on the wall there for some time).
 
Not really, I heavily doubt that Folding@Home is latency sensitive in anyway.

It's cloud computing, so yes. I also didn't say they were exactly the same. Your right about the latency part, Folding@Home was app to use while your ps3 was idle.

It's not hard to have a powerful server with 16 virtual machines running. One VM acting solely as a server while the other 15 VMs acting as thin clients (ps3). Connecting them via a virtual star topology sharing a virtual router. BOOM

getting my CCNA right now and we covered this.
 
They are currently using this in a AAA title or in the next year globally?

You guys seem to be confusing whether the technology exists for it and whether or not these companies can afford to utilize it.

It all depends on applications and costs and Microsoft is in a position to offer it while most couldn't or wouldn't risk it.

What puts Microsoft in any better of a position than any AWS customer? Yes, they own Azure, but every server they dedicate to gaming is a loss to a paying business customer.
 
The sheer power of the Xbox One infrastructure is finally here. I have both consoles but you have to hand it to Microsoft, the real "hidden Gpu" secret sauce was the cloud. I hope Sony takes notes before Microsoft figures out how to offload the minuscule resolution problem to the cloud too!
 
Its a mess in here.


Im all for this, Especially in a game like Crackdown but I always have to think that as more games get released that may partially require services in order for certain portions to work properly, what will that do for game conservation?

We know there are dozens of games that need online services and they can't run forever... What happens when that time is done? Nobody can play that mode, maybe even the actual full game again?

If I'm not mistaken, with Azure, won't it just be able to save off the specific configuration for that particular game? So games like Titanfall etc. potentially have access to vm servers at any time, but if they're not in use, they just get spun down and the resources put back into the pool for other games to be able to use. So all they have to keep is that configuration available to use when a vm needs to be spun up.
 
7
It's cloud computing, so yes. I also didn't say they were exactly the same. Your right about the latency part, Folding@Home was app to use while your ps3 was idle.

It's not hard to have a powerful server with 16 virtual machines running. One VM acting solely as a server while the other 15 VMs acting as thin clients (ps3). Connecting them via a virtual star topology sharing a virtual router. BOOM

getting my CCNA right now and we covered this.

What I'm saying is that cloud computing isn't hard or new; cloud computing for real time physics calculations for 99 people on the same server is.

Traditionally in a P2P setting a multiplayer games performance is limited by the power of the host hardware. Crackdown 3 is basically ballooning the host hardware by 20x and running the physics calculations for the session.

What puts Microsoft in any better of a position than any AWS customer? Yes, they own Azure, but every server they dedicate to gaming is a loss to a paying business customer.

Azure servers aren't dedicated to anything; a VM can go from hosting a Forza Motorsport 5 race to streaming a Office 365 client.

What puts the second largest server company(largest server renting company) in a better position than a AWS customer? I wonder...
 
Played this at Gamescom after a presentation by the team, I was seriously in awe by the physics involved. We took down a tall building in order to take down the bridge between two other structures next to it, not a single dropped frame.

I'v been very skeptical about cloud computing, but I can't argue that what I saw and played was very impressive.

I suppose you could fill in all the "I'll believe it when I see it/when you can play it" people.

Heheh
 
Crackdown's destruction is insane. I don't know how people can compare it to other games, there's no game that compares to the level of destruction shown in the video.

Exciting times ahead and I'm looking forward to see how much they push the cloud in games. I remember Kampfheld posting grass spline gif's, so it's not just destruction it can be used for. I hope to see more games take advantage of it in different ways especially if the outcome is anything close to as big as the CD destruction.

Speaking of Kampfheld will he ever be unbanned? Would be interesting to hear him out now the CD demo is released and everything he said about CD lined up so maybe he was working on it after all.

I've posted this before, but Bish banned him because he said he was working on Crackdown. DaveJ said he wasn't, but DaveJ is really only in charge of Reagent and Cloudgine and not Sumo. It sounded more like a cover the ass post from him to get Kamp to shut up and Kamp never stated he worked for Cloudgine. 95% of what he posted was right on the money including how they were thinking about splitting the game up. He switched jobs last year, and I don't see how someone could pull a long con for that long, and magically have inside information and comment that he wasn't working at a studio that everyone thought was working on the game.
 

It's very cool.

I'm interested to learn what it takes to bring down buildings with normal-power weapons. I imagine that they'll try to keep the cloud-based computations to a reasonable level by making the buildings a little more resilient than what we're seeing.

I'm not complaining at the prospect of strong buildings, that just means it's going to take team work and a bit of creativity to achieve big destruction.

The sheer power of the Xbox One infrastructure is finally here. I have both consoles but you have to hand it to Microsoft, the real "hidden Gpu" secret sauce was the cloud. I hope Sony takes notes before Microsoft figures out how to offload the minuscule resolution problem to the cloud too!

I think by the time this matures to a level that it could pull of some of the things they were suggesting back at Xbox One launch we'll be in another generation of consoles, or they may move straight to running games in the cloud.

I've posted this before, but Bish banned him because he said he was working on Crackdown. DaveJ said he wasn't, but DaveJ is really only in charge of Reagent and Cloudgine and not Sumo. It sounded more like a cover the ass post from him to get Kamp to shut up and Kamp never stated he worked for Cloudgine. 95% of what he posted was right on the money including how they were thinking about splitting the game up. He switched jobs last year, and I don't see how someone could pull a long con for that long, and magically have inside information and comment that he wasn't working at a studio that everyone thought was working on the game.

He was wrong about Banjo being at E3.

I know things can change quickly in the industry, but he seemed so certain it was happening.
 
Not sure where there was any mention of games that do this ("AAA" or otherwise), obviously nothing is on the market yet. And if Crackdown is first (which is likely) we're still 1-2 years out from seeing it in action over "NA/Europe & Asia".


Gaf always trivializes any engineering accomplishments, but that's besides the point. If you ever looked at online development and cost-margins there, the notion that "cloud power" is free in any sense of the word is more than a little absurd (and I'm purely talking compute $ here, nothing related to complexity or ease of utilizing it).
The idea of running large scale game simulations entirely server-side is nice, but it pretty much spells the end of "pay for product" business model IMO (then again the writting has been on the wall there for some time).

I wasn't arguing the economics, and understand that any server infrastructure, even if owned, wouldn't be free (or even cheap). I was merely addressing the point that "Sony can do this too! Look, they have servers!"...ok, but that misses the entire point about what's being done on the software side of things, which is not just a matter of having a bunch of hardware hooked up. There's more to it than just the servers.

In fact, I'd argue that acquiring the servers is easier/more straightforward part of the process. At least that is well understood - the software of making this all come together in a game is completely new, and not something that can just be done overnight.
 
What I'm saying is that cloud computing isn't hard or new; cloud computing for real time physics calculations for 99 people on the same server is.

It's really not. The model of the building in an instance's memory and calculations are applied to the model based on input from players and then streamed back to all players. The map is sharded across multiple instances. There are numerous cloud applications that are more complex than this.
 
What puts Microsoft in any better of a position than any AWS customer? Yes, they own Azure, but every server they dedicate to gaming is a loss to a paying business customer.

I think people may be underestimating the amount of software development work it takes to do something like this. Anything requiring low-latency over the network, with parallel processing among multiple servers is a ton of work. It's not easy to do, so you need someone like Microsoft (or Sony or maybe a publisher as big as EA) to champion this for the game developers.

In theory anyone could do this. It's just that Microsoft may have a couple years worth of useful libraries, components, and a server stack that can do a lot of the work for the game developers already built.
 
So far, Crackdown 3 is the only next gen title that has truly impressed me and made me feel like this is "next gen".
 
I'm a bit of a Sony/PS4 fan of late, definitely a fanboy of Sony if you will, but I'm not blind to very impressive cool tech like this. It does look seriously impressive. The guy did state this was a tech demo of what could be done rather than an exact look at how this tech will manifest itself in the actual game. I imagine destruction like this could break the gameplay in many instances, it's definitely something that will need to be implemented in a clever way.

As for the raw tech itself.. there must be some drawback..? Other than the obvious of needing I presume a decent internet connection. Surely all of this data being done away from the XB1 still needs to get sent. Every last grain of concrete created from the destruction still needs to have data about it sent down the pipeline so it can be rendered where it's supposed to be, and surely if you go full nuclear in the game it must lead to some bottlenecks somewhere...? If not then fair play to them!
 
They are currently using this in a AAA title or in the next year globally?

You guys seem to be confusing whether the technology exists for it and whether or not these companies can afford to utilize it.(If the infrastructure exists in a way to offer the services globally too.)

It all depends on applications and costs and Microsoft is in a position to offer it while most couldn't or wouldn't risk it.

It isn't being done now, in this one tightly constrained definition I'm using to advance my point, so it can't be done ever is some seriously screwed up logic.
 
It's really not. The model of the building in an instance's memory and calculations are applied to the model based on input from players and then streamed back to all players. The map is sharded across multiple instances. There are numerous cloud applications that are more complex than this.

Oh my god......... we are not talking about numerous cloud applications.

We are talking about gaming.

That is latency sensitive.

Being offered globally.

Everything you keep saying just sounds like it is trying to downplay this accomplishment.

Show me a single other game on any platform that is matching the experience Crackdown 3 is offering in the availability it is offering it in.

It isn't being done now, in this one tightly constrained definition I'm using to advance my point, so it can't be done ever is some seriously screwed up logic.

It's a good thing I never said that then, isn't it?
 
I suppose you could fill in all the "I'll believe it when I see it/when you can play it" people.

Heheh
Not much else I can do at the moment, writing up my impressions though. It was one of the few titles that I've seen this year that actually seemed like it lived up to what they were aiming for, technically anyway. I do share concerns about cloud computing when it comes to player-bases though. At one point the screen noted about 10 cloud servers were being used, and even though it looked impressive, the fact is still that it's all lost if MS decides to turn off the servers, pretty much crossing off one of the biggest potential selling points of the game.

I'd love to see a demo across an international network though, especially with varying connection speeds, since for demo purposes they understandably opted for an optimal network for PR purposes.
 
It's too bad Crackdown 3 still has that cel-shade look because I think it looks awful. :/

I think it looks great in the higher res IGN video I just posted, but yeah, if you don't like the art style, you don't like the art style. I think it lends itself well to this technology.
 
It's very cool.

I'm interested to learn what it takes to bring down buildings with normal-power weapons. I imagine that they'll try to keep the cloud-based computations to a reasonable level by making the buildings a little more resilient than what we're seeing.

I'm not complaining at the prospect of strong buildings, that just means it's going to take team work and a bit of creativity to achieve big destruction.



I think by the time this matures to a level that it could pull of some of the things they were suggesting back at Xbox One launch we'll be in another generation of consoles, or they may move straight to running games in the cloud.



He was wrong about Banjo being at E3.

I know things can change quickly in the industry, but he seemed so certain it was happening.

He said he didn't work at Rare, but we got Rare Replay?
 
Those companies also only offer those kind of services by regions and only support those regions based on what can be provided.

Microsoft has to provide it to every region they game will be available in and have mostly the same experience in every region.

You can't compare 3D modeling cloud compute(which isn't even latency sensitive) to streaming physics computations in a latency sensitive setting.

One of the main reasons to use existing cloud infrastructure is to provide your services in multiple regions. For example, I can provision VM's in Azure or any other cloud provider in all the regions they offer. That would also have a lot to do with latency, along with programming/API integration (which I've already said MS is far more advanced in compared to any other game developers).

Whether it's cost effective would require a lot more information on the compute requirements of the particular game vs cost of renting infrastructure, but arguing it's simply impossible is ridiculous. This is especially true as cloud services expand out even more and start including GPU compute (I believe AWS is already doing this).
 
I've posted this before, but Bish banned him because he said he was working on Crackdown. DaveJ said he wasn't, but DaveJ is really only in charge of Reagent and Cloudgine and not Sumo. It sounded more like a cover the ass post from him to get Kamp to shut up and Kamp never stated he worked for Cloudgine. 95% of what he posted was right on the money including how they were thinking about splitting the game up. He switched jobs last year, and I don't see how someone could pull a long con for that long, and magically have inside information and comment that he wasn't working at a studio that everyone thought was working on the game.

Yea I know what went down, I thought it was a bit harsh since he always said he was working at the company developing the game and that it wasn't announced yet but wasn't cloudgine, which I think was new information at the time.

I understand why he was banned even if it was a bit harsh and in a way he did ask for it with the "Ban me until you see at E3 if you want" line, but now that CD demo is out and we can see he was spot on for pretty much everything, it would be nice to have him back here.
He was wrong about Banjo being at E3.

I know things can change quickly in the industry, but he seemed so certain it was happening.
Banjo was at E3, it just wasn't a new Banjo

I know most people took it as if it was a new banjo, myself included, but he never said that.
 
I wonder if MS released a demo video that showed the attainable differences between using the cloud in game and not using it, if it would curb some of the scepticism.

Also, I wonder how long it will take the executives at MS to eventually shoot themselves in the foot, by licensing out the tech to their competitors. They have a nice habit of doing this.
 
Top Bottom