As technology ages, it gets harder to maintain, though.
I mean.. there's going to come a day when we leave the x86 architecture behind. It's just GOING to happen. May take 20 years, may take the Crash 1.0, but it will happen. And after that point, what then?
x86 is just a way of encoding and decoding instructions now and has very few ties to the actual implementation beyond the bit of silicon that decodes them. This exact argument was floated about 20 years ago too, and also more recently with the 32/64-bit stuff and Itanium until AMD finished up with their version.
x86 contains a ton of cruft sure, and there are architectures that are better, however there's no incentive to ever switch to anything else. Itanium is basically better than x86, but the compilers we needed were too hard to write properly. ARM/MIPS/RISC have problems of their own. x86 instructions can be reclaimed and repurposed as needed and this has already been done in the initial 64-bit implementation.
If we get to the point where all the apps are running in data centers and accessed through phones/web, then I could see it maybe happening there, but there's too much expectation of compatibility in the consumer market
The Division, BF4 and DR3 will all be using Smartglass technology to affect someone playing on the console. That's not possible without the cloud.
I don't know why you responded to a post about processor architectures with cloud buzzwords, but from what I can tell smartglass itself could be done over local wifi just as well
I responded too "If we get to the point where all the apps are running in data centers and accessed through phones/web, then I could see it maybe happening there, but there's too much expectation of compatibility in the consumer market".
You can but it's is all run on dedicated hardware that resides in the cloud. So its available everywhere. You could play commando mode on the bus on the way to work if you wanted too. Again, it's all things Sony or Nintendo can do but they have to build their own infrastructure or lease cloud computing to do it where Microsoft offers it as an integral part of the Xbox live network. The company I work for (not MS in case someone ask) is building huge data center base on the same premise. Everything is running in VM. Its not longer your typical physical server per application but array of processing cycle you can use on demand.
0
Options
AegeriTiny wee bacteriumsPlateau of LengRegistered Userregular
The problem is, which actually I didn't even consider at first, is you forgot to factor in latency for a multiplayer game like BF4. If your commander is well behind on the game and lagging compared to people actually playing then that's not very good. "Oh, you threw away your tomahawk missile at a tank that had already moved away, um.. okay" is not going to endear any commander to those actually playing the game. This makes me immediately realize azure or microsofts cloud is probably going nowhere near this in terms of actually using it for this purpose. So this discussion really doesn't have much of a point, unless EA/Ubisoft want to hamstring their games functionality by making them even more at the mercy of latency than they already are :P
The problem is, which actually I didn't even consider at first, is you forgot to factor in latency for a multiplayer game like BF4. If your commander is well behind on the game and lagging compared to people actually playing then that's not very good. "Oh, you threw away your tomahawk missile at a tank that had already moved away, um.. okay" is not going to endear any commander to those actually playing the game. This makes me immediately realize azure or microsofts cloud is probably going nowhere near this in terms of actually using it for this purpose. So this discussion really doesn't have much of a point, unless EA/Ubisoft want to hamstring their games functionality by making them even more at the mercy of latency than they already are :P
I guess we will know soon enough. I am curious to see how that mode will work. If they spend the money to develop this mode, it must somehow work in a near acceptable latency perspective. Although you could be an 100% right, it could render the mode unusable.
*Edit
Its not harder to sync a tablet/phone than a console at least on Wifi and LTE network.
Symtex on
0
Options
AegeriTiny wee bacteriumsPlateau of LengRegistered Userregular
edited June 2013
My point is that I am betting they just connect to exactly the same dedicated server that everyone else is playing on. So it probably has the same latency as what everyone else on the server has (as these servers are handling up to 64 people anyway, 66, two of whom are entirely on a different device needing different things, isn't a huge stretch). It will be interesting how they handle it, but I sincerely doubt reading the descriptions of commander mode that it does anything other than use the same dedicated servers. Kind of like having 64 player slots and then 2 for a commander (or possibly the commander takes up a player slot? Not sure).
Either way, I am far from convinced this is a relevant or even desirable application for many FPS or multiplayer games that are latency dependant.
The best case I can see for it is just to run the usual multiplayer servers and such. Part of the problem with relying on a server farm is what happens on release day when literally millions of people start playing a game at the same time that's relying on the cloud for critical processing? The whole "power of the cloud" is somewhat constrained by the fact that you're sharing it with everyone else and whatever other obligations MS has. What the cloud is good for is:
- shuffling data around, savefiles and such
- multiplayer servers and the like
- burst processing. They can easily throw a thousand cores at something for you very quickly, as long as you only need them for a second or two every hour
I'm on the fence with procedurally generated maps. For something like a strategy game, with random encounters? I could see it working, so long as you got the tech locked down so it never creates anything weird. Open world games though? The hand crafted world is always going to have infinitely more personality than the RNG one. Plus it will generally make the dev's lives easier when it comes to creating cutscenes and story bits and scripted sequences and the like. If the world is random, then you have to be general in how you do things, because the world is always going to be different, so your scenarios have to take that into account.
Ultimately I remain highly skeptical on cloud computing. Most of the examples given so far can already be done. A lot of the cooler ones don't sound feasible with current latency issues. And a few sound cool and possible... but it's also far easier to just fake the damn thing, because the player never knows the difference.
"The sausage of Green Earth explodes with flavor like the cannon of culinary delight."
Ah, that reminds me of something I was thinking. In the long term, the biggest difference cloud computing could make is to make games more "real". Right almost all games are nothing but Potemkin villiages. It's all fascade and fakery with nothing behind it all. That's true of AI and physics and the the long, long corridors we walk down to get to the credits. Add in enough computing power and less and less would have to be faked and fewer limitations put on the player.
At the same time, the way games would be made would be totally different.
(Please do not gift. My game bank is already full.)
0
Options
syndalisGetting ClassyOn the WallRegistered User, Loves Apple Productsregular
Ah, that reminds me of something I was thinking. In the long term, the biggest difference cloud computing could make is to make games more "real". Right almost all games are nothing but Potemkin villiages. It's all fascade and fakery with nothing behind it all. That's true of AI and physics and the the long, long corridors we walk down to get to the credits. Add in enough computing power and less and less would have to be faked and fewer limitations put on the player.
At the same time, the way games would be made would be totally different.
I've been paying a lot of attention to Titanfall, as it is one of the first games that seems to be pushing "cloud" rather heavily.
It lends itself to your point about how games will be made differently, though. Supposedly the maps are divorced from the scenarios. At any point a storied "scenario" can crop up in your co-op multiplayer experience, which will carry itself over multiple maps and stages. These scenarios adapt to the conditions, the maps, the number of players, etc. And can provide a couple hours of compelling, story-driven gameplay that won't be the same the next time you encounter it.
That seems kind of neat.
SW-4158-3990-6116
Let's play Mario Kart or something...
0
Options
Giggles_FunsworthBlight on DiscourseBay Area SprawlRegistered Userregular
People interested in procedural generation should check out .kkrieger. It's a 96kb fully 3D FPS that doesn't look like total shit, especially considering it's almost 10 years old now.
Microsoft has said that there are 300,000 servers in Azure to serve the VM/authentication/computing needs of the Xbone players. They have also said that "the cloud" will give each Xbone the power of three additional Xbones to give you an experience that the cloudless lolsony PS4 can't provide.
Latency aside. Bandwidth aside. Bullshit PR game-enhancing promises that are blatant lies aside.
How is Microsoft going to handle, let alone afford, the "One BILLION console!!" market they are trying to grasp in their iron mailed gauntlet?
Let's say, just for shits and giggles, that the entire 300,000 server infrastructure that makes up Azure is used solely for Xbone cloud whatever. This excludes every enterprise application, every private user, lab, everything (which I would assume is a much larger user than what we would believe, as well as a much higher paying customer) that is not Xbone related.
Even if Microsoft launches their Xbone and only sells 500,000 systems (which is a horribly low number for what they're trying to accomplish), each of those 300,000 servers would have to be MASSIVELY powerful PCs to make up for the "3 Xbones in the cloud to work for your Xbone" claim.
Would it be a conservative estimate that each of those servers would have to contain at least $3,000 worth of enterprise-grade parts to even come close to meeting the projection of supplying the initial launch of 500,000? Is it even possible to build each of those 300,000 servers with enough high-powered parts to equal 1.5 million Xbones? We're talking that ideally, each server would have to be as powerful as 5 Xbones to meet the "3 Xbones for every sold Xbone" on 500,000 Xbone sales .
Even if I were to build a personal computer that had the processing power of 5 Xbones, it would cost me over $3,000, and that's with PC parts, which are a fraction of the price of enterprise-grade parts. I have no idea if Microsoft's ability to purchase huge orders of Dell servers counterbalance the enterprise level pricing, so this is just a completely uninformed assumption on my part. Please take this with a Dead Sea sized grain of salt.
Each of those shipping crate sized server houses hold anywhere from 1,800-2,500 servers. At $3,000 each, you're looking at $5.4 million-$7.5 million per server house.
For 300,000 servers, you're looking at a total of $900 million.
Even though they can add on additional server houses as needed, each 2,500 maxed out server house would cost an additional $7.5 million dollars to add onto the stack.
All of this is supposed to be subsidized with XBL? Can Dell (or any other company) pump out thousands of Azure blades each month to meet demand?
This makes no sense whatsoever. To meet what Microsoft's boasts of the power of the cloud, each of those 300,000 server blades would have to be Cray level supercomputers. Hell of a deal for my completely hypothetical assumption of $3,000 each.
Again, this doesn't include every company, every lab, every single other customer that Azure was originally built for, all of which use gargantuan levels of space and computing power for their own purposes. These 300,000 servers are exclusively for Xbone use only.
What happens if they sell a million Xbones? 10 million? 100 million? All of these numbers are horrendously lower than what Microsoft is projecting if they're planning on making money on all of their world-conquering plans, and yet, it seems like the cloud isn't big enough to accommodate even 1/100th of the Xbones Microsoft wants up and running if they're going to make money.
You can't make the argument that "not all Xbone players will be on simultaneously, so they don't need to have the servers available at all times" because, yeah, all Xbones will always be connected. I highly doubt that Azure will be programmed to disconnect a Forbes 500 company's VM to make space for your CoD multiplayer game, nor will they stop some cancer lab's continually running gene-crunching program so you can be too lazy to drive your own video game car.
There's way too many things that make absolutely no sense whatsoever. There's too many "what-ifs", and Microsoft has been going out of their way to not say anything outside alternating between shooting themselves in the foot and espousing dreams made of unicorn love and ice cream sex.
Seriously, in my own personal opinion, "the cloud" will have to do very, very little, if anything at all, to accommodate the number of Xbones Microsoft wants connected to the cloud. It would be absolutely ridiculous to assume that these server blades are capable of computing on a level to "enhance" the game through sheer computing while simultaneously accommodating all the needs of the enterprise-level accounts who pay shitloads of money to utilize Azure.
I'm sure there will be cloud saves, leaderboards, the occasional MMO-type server-side programs, and of course the authentication pinging, but for there to be serious amounts of number crunching for a game seems highly doubtful to me, especially when big, new games come out and there's a million people trying to create a VM for their deathmatch map.
Am I wrong? Am I completely off? Do I not have a clue what computers and clouds are?
I'm not sure your numbers or assumptions are right (for instance the cloud is exactly designed to switch from deathmatch servers to some random corporate use). But I too am very interested in the economics behind providing large numbers of CPU cycles to large numbers of consoles. If someone knows more I'd love to hear it.
As someone else mentioned, the biggest advantage is in using a large number of CPUs to do some calculation very fast and then releasing all those assets so that the total cost is low.
IMO the high power of the XBO makes it harder to leverage the cloud in this way.
(Please do not gift. My game bank is already full.)
0
Options
HenroidMexican kicked from Immigration ThreadCentrism is Racism :3Registered Userregular
A big problem I have with the concept of cloud computing (from the angle of handling physics / AI / graphics) is that those processes take time to process. The synchronization problems this can present must be every programmer's nightmare. You would have to create some very, very, very static conditions in the game in order for pre-loading to be viable. That seems impossible to me when it comes to lighting, which is one of the biggest drags in graphics processing while also having the demand of being real time and dynamic.
They are definitely hoping that they can have the servers on yet have people only play a few hours a day and so the peak users at any given moment is low enough that they can handle. Though with their US focus that won't work quite as well as a worldwide release! And launch days are going to be horrible for it
No, cloud computing isn't the future of gaming. Cloud computing is a buzzword being used to try to take control of your data, your privacy out of your control.
Governments wants cloud computing sooo bad so they can snoop even more of our data. Corporations want cloud computing so bad so they can buy/sell/trade your data with other entities (including government)
I believe the GNU creator, that cloud computing is a trojan horse to get your data out of your control and away from your hands. Cloud computing is a way for companies to monetize services that should be free that you take for granted on your own devices.
i don't think gaming will ever really benefit from cloud computing... besides just the standard fair we have seen over the last 10 years, like centralized servers serving up multiplayer game content (WoW, FPS)
Each of those shipping crate sized server houses hold anywhere from 1,800-2,500 servers. At $3,000 each, you're looking at $5.4 million-$7.5 million per server house.
For 300,000 servers, you're looking at a total of $900 million.
Even though they can add on additional server houses as needed, each 2,500 maxed out server house would cost an additional $7.5 million dollars to add onto the stack.
All of this is supposed to be subsidized with XBL?
No, all of this is subsidized with Office. Putting XBL on Azure means that their hardware is still being heavily utilized outside office hours. They may be expanding Azure, but the idea that they're doing it solely for XBL is purest spin.
Hmmmm, I stumbled onto this thing. Seems like a Azure server that might be up to the power of the XBO cost about $.40 per hour. If, say, MS was making 3 of those available for each XBO per their PR and you played a 8 hour game that would be Something like $10 in cloud costs for your playthrough.
Now I don't know if that should be discounted because of non-peak rates. Then again, 50 million XBO might turn work hours into the non-peak times for azure.
Just doing that rough calculation makes me wonder if MS can realistically provide significant computing power with Azure if the XBO allows used games. The could would turn the situation of not getting any money on used games to one where MS (or publishers if they are charged) lose money on used games.
Of course none of this means dedicated servers or cloud saves or whatever would be uneconomic. I'm just talking about significantly increasing the computing power of the XBO.
(Please do not gift. My game bank is already full.)
0
Options
syndalisGetting ClassyOn the WallRegistered User, Loves Apple Productsregular
Hmmmm, I stumbled onto this thing. Seems like a Azure server that might be up to the power of the XBO cost about $.40 per hour. If, say, MS was making 3 of those available for each XBO per their PR and you played a 8 hour game that would be Something like $10 in cloud costs for your playthrough.
Now I don't know if that should be discounted because of non-peak rates. Then again, 50 million XBO might turn work hours into the non-peak times for azure.
Just doing that rough calculation makes me wonder if MS can realistically provide significant computing power with Azure if the XBO allows used games. The could would turn the situation of not getting any money on used games to one where MS (or publishers if they are charged) lose money on used games.
Of course none of this means dedicated servers or cloud saves or whatever would be uneconomic. I'm just talking about significantly increasing the computing power of the XBO.
the actual cost is much much lower than what they charge outside people to rent their machines.
We aren't getting Microsoft's internal cost metrics because why would we?
Lets just say that the 60/yr people pay for live gold will likely break even the expense of running Azure for them, if not net a small profit, as it value adds to their service and convinces people to purchase more games, especially through the digital distribution service if they can really nail that.
SW-4158-3990-6116
Let's play Mario Kart or something...
Galactic Reign is a Microsoft game that was co-developed with Canadian developer Slant Six Games that pretty much came out of nowhere when it appeared on Xbox Windows Phone 7 and 8 and Windows 8 and RT last week.
...
The really unusual aspect of the battles is that they are computed entirely in the cloud as opposed to the game client. And rather than an in-game engine, the game renders the battle in the cloud. The results are then streamed to your device as a video. Players can choose to ‘scan’ the video, which adds tactical overlays and allows you to jump to any point in the battle.
Might be cool, no?
Problem is, it's shutting down at the end of this year.
MS is dumb for even letting something like this happen as they try to launch a new console with "infinite power of the cloud" as one of the talking points.
(Please do not gift. My game bank is already full.)
I'd tried it out a couple times -- it's an asynchronous multiplayer game (there's barely any content that can be done solo, mostly just to teach you the interface) where you and your opponent both create customized space armadas and see who wins in a fight. It was a pretty neat game, but I just didn't really have much interest in that type of gameplay to try it beyond that.
Galactic Reign is a Microsoft game that was co-developed with Canadian developer Slant Six Games that pretty much came out of nowhere when it appeared on Xbox Windows Phone 7 and 8 and Windows 8 and RT last week.
...
The really unusual aspect of the battles is that they are computed entirely in the cloud as opposed to the game client. And rather than an in-game engine, the game renders the battle in the cloud. The results are then streamed to your device as a video. Players can choose to ‘scan’ the video, which adds tactical overlays and allows you to jump to any point in the battle.
Might be cool, no?
Problem is, it's shutting down at the end of this year.
MS is dumb for even letting something like this happen as they try to launch a new console with "infinite power of the cloud" as one of the talking points.
But cloud games shouldn't ever need to be shut down. Whenever the cloud is needed, a VM will quickly be generated to host that game briefly, now and for the foreseeable future, right?
AegeriTiny wee bacteriumsPlateau of LengRegistered Userregular
edited August 2013
Remember all those people who were insistent with me that "The cloud" would mean servers would be practically free (they aren't) and so the main advantage was that server support for azure games would last forever? Of course this was entirely unlike the clearly bad and backwards idea of just making dedicated servers doing exactly the same things, but not attaching a meaningless marketing buzzword to it. What's the difference again? Other than cost, where Azure virtual server space is less costly than a traditional dedicated server - which I think at this point is all we are left with on the "What does this actually do to benefit anyone?" question. Just like any dedicated server model though, one it becomes not worthwhile to keep servers for a game up they get shut down and this game proves precisely that.
The interesting thing for this thread is that he expresses total panic at the idea that the code for their games actually be run in parallel outside of some special circumstances (basically completely trivial cases, something like there are X particles, do Y to each but not letting the particles interact). In fact he suggests an complete different programming paradigm would be needed. He mused about how perhaps he should have made up QuakeScheme instead of QuakeC then that might have made things easier because basically the little mistakes that C or C++ let you make would start freezing the game at random times. He's actually talking about using multiple cores more efficiently but I think it's interesting for cloud computing in that cloud computing has basically the same problem but it's even worse because it's less predictable. Anyway the extremity of his solution might convey the level of difficulty in writing programs that both run in parallel and interact with the user.
On the other hand he is a big proponent of the idea that having more computing power available will make things simpler for both programmers (which is obvious) and content creators (not as obvious but basically less tricks are needed when you are rendering things more realistically). And that of course is part of the long term benefit of cloud computing as we discusses a while back.
(Please do not gift. My game bank is already full.)
Microsoft has said that there are 300,000 servers in Azure to serve the VM/authentication/computing needs of the Xbone players. They have also said that "the cloud" will give each Xbone the power of three additional Xbones to give you an experience that the cloudless lolsony PS4 can't provide.
Latency aside. Bandwidth aside. Bullshit PR game-enhancing promises that are blatant lies aside.
How is Microsoft going to handle, let alone afford, the "One BILLION console!!" market they are trying to grasp in their iron mailed gauntlet?
Let's say, just for shits and giggles, that the entire 300,000 server infrastructure that makes up Azure is used solely for Xbone cloud whatever. This excludes every enterprise application, every private user, lab, everything (which I would assume is a much larger user than what we would believe, as well as a much higher paying customer) that is not Xbone related.
Even if Microsoft launches their Xbone and only sells 500,000 systems (which is a horribly low number for what they're trying to accomplish), each of those 300,000 servers would have to be MASSIVELY powerful PCs to make up for the "3 Xbones in the cloud to work for your Xbone" claim.
Would it be a conservative estimate that each of those servers would have to contain at least $3,000 worth of enterprise-grade parts to even come close to meeting the projection of supplying the initial launch of 500,000? Is it even possible to build each of those 300,000 servers with enough high-powered parts to equal 1.5 million Xbones? We're talking that ideally, each server would have to be as powerful as 5 Xbones to meet the "3 Xbones for every sold Xbone" on 500,000 Xbone sales .
Even if I were to build a personal computer that had the processing power of 5 Xbones, it would cost me over $3,000, and that's with PC parts, which are a fraction of the price of enterprise-grade parts. I have no idea if Microsoft's ability to purchase huge orders of Dell servers counterbalance the enterprise level pricing, so this is just a completely uninformed assumption on my part. Please take this with a Dead Sea sized grain of salt.
Each of those shipping crate sized server houses hold anywhere from 1,800-2,500 servers. At $3,000 each, you're looking at $5.4 million-$7.5 million per server house.
For 300,000 servers, you're looking at a total of $900 million.
Even though they can add on additional server houses as needed, each 2,500 maxed out server house would cost an additional $7.5 million dollars to add onto the stack.
All of this is supposed to be subsidized with XBL? Can Dell (or any other company) pump out thousands of Azure blades each month to meet demand?
This makes no sense whatsoever. To meet what Microsoft's boasts of the power of the cloud, each of those 300,000 server blades would have to be Cray level supercomputers. Hell of a deal for my completely hypothetical assumption of $3,000 each.
Again, this doesn't include every company, every lab, every single other customer that Azure was originally built for, all of which use gargantuan levels of space and computing power for their own purposes. These 300,000 servers are exclusively for Xbone use only.
What happens if they sell a million Xbones? 10 million? 100 million? All of these numbers are horrendously lower than what Microsoft is projecting if they're planning on making money on all of their world-conquering plans, and yet, it seems like the cloud isn't big enough to accommodate even 1/100th of the Xbones Microsoft wants up and running if they're going to make money.
You can't make the argument that "not all Xbone players will be on simultaneously, so they don't need to have the servers available at all times" because, yeah, all Xbones will always be connected. I highly doubt that Azure will be programmed to disconnect a Forbes 500 company's VM to make space for your CoD multiplayer game, nor will they stop some cancer lab's continually running gene-crunching program so you can be too lazy to drive your own video game car.
There's way too many things that make absolutely no sense whatsoever. There's too many "what-ifs", and Microsoft has been going out of their way to not say anything outside alternating between shooting themselves in the foot and espousing dreams made of unicorn love and ice cream sex.
Seriously, in my own personal opinion, "the cloud" will have to do very, very little, if anything at all, to accommodate the number of Xbones Microsoft wants connected to the cloud. It would be absolutely ridiculous to assume that these server blades are capable of computing on a level to "enhance" the game through sheer computing while simultaneously accommodating all the needs of the enterprise-level accounts who pay shitloads of money to utilize Azure.
I'm sure there will be cloud saves, leaderboards, the occasional MMO-type server-side programs, and of course the authentication pinging, but for there to be serious amounts of number crunching for a game seems highly doubtful to me, especially when big, new games come out and there's a million people trying to create a VM for their deathmatch map.
Am I wrong? Am I completely off? Do I not have a clue what computers and clouds are?
Your concerns are valid, but at the same time you're assuming that Azure is something that was created specifically for XBone. It already exists and processes data for close to a billion users daily.
You're also making the assumption that every user around the world is using all the bandwidth at the same time.
I don't believe the claim that the cloud will triple the power of an Xbone. But I'm pretty confident that Asure has the infrastructure to handle the real world demands of what Microsoft can expect.
Edit - woah oooold draft slipped in there, deleted now
No, cloud computing isn't the future of gaming. Cloud computing is a buzzword being used to try to take control of your data, your privacy out of your control.
Governments wants cloud computing sooo bad so they can snoop even more of our data. Corporations want cloud computing so bad so they can buy/sell/trade your data with other entities (including government)
I believe the GNU creator, that cloud computing is a trojan horse to get your data out of your control and away from your hands. Cloud computing is a way for companies to monetize services that should be free that you take for granted on your own devices.
i don't think gaming will ever really benefit from cloud computing... besides just the standard fair we have seen over the last 10 years, like centralized servers serving up multiplayer game content (WoW, FPS)
Sounds about right, maybe a touch on the paranoid side, but basically it's not really about efficiency. If it's so expensive to host server farms and bloo bloo bloo look at the size of this cross we must bear, then why are they so eager to make video games (like simcity) that could have simply been single play local games into online monstrosities? Not out of generosity or desire to deliver more features.
Posts
I responded too "If we get to the point where all the apps are running in data centers and accessed through phones/web, then I could see it maybe happening there, but there's too much expectation of compatibility in the consumer market".
You can but it's is all run on dedicated hardware that resides in the cloud. So its available everywhere. You could play commando mode on the bus on the way to work if you wanted too. Again, it's all things Sony or Nintendo can do but they have to build their own infrastructure or lease cloud computing to do it where Microsoft offers it as an integral part of the Xbox live network. The company I work for (not MS in case someone ask) is building huge data center base on the same premise. Everything is running in VM. Its not longer your typical physical server per application but array of processing cycle you can use on demand.
I guess we will know soon enough. I am curious to see how that mode will work. If they spend the money to develop this mode, it must somehow work in a near acceptable latency perspective. Although you could be an 100% right, it could render the mode unusable.
*Edit
Its not harder to sync a tablet/phone than a console at least on Wifi and LTE network.
Either way, I am far from convinced this is a relevant or even desirable application for many FPS or multiplayer games that are latency dependant.
- shuffling data around, savefiles and such
- multiplayer servers and the like
- burst processing. They can easily throw a thousand cores at something for you very quickly, as long as you only need them for a second or two every hour
Ultimately I remain highly skeptical on cloud computing. Most of the examples given so far can already be done. A lot of the cooler ones don't sound feasible with current latency issues. And a few sound cool and possible... but it's also far easier to just fake the damn thing, because the player never knows the difference.
Ah, that reminds me of something I was thinking. In the long term, the biggest difference cloud computing could make is to make games more "real". Right almost all games are nothing but Potemkin villiages. It's all fascade and fakery with nothing behind it all. That's true of AI and physics and the the long, long corridors we walk down to get to the credits. Add in enough computing power and less and less would have to be faked and fewer limitations put on the player.
At the same time, the way games would be made would be totally different.
(Please do not gift. My game bank is already full.)
I've been paying a lot of attention to Titanfall, as it is one of the first games that seems to be pushing "cloud" rather heavily.
It lends itself to your point about how games will be made differently, though. Supposedly the maps are divorced from the scenarios. At any point a storied "scenario" can crop up in your co-op multiplayer experience, which will carry itself over multiple maps and stages. These scenarios adapt to the conditions, the maps, the number of players, etc. And can provide a couple hours of compelling, story-driven gameplay that won't be the same the next time you encounter it.
That seems kind of neat.
Let's play Mario Kart or something...
http://www.gamasutra.com/view/feature/2277/interview_frugal_fragging_with_.php
Microsoft has said that there are 300,000 servers in Azure to serve the VM/authentication/computing needs of the Xbone players. They have also said that "the cloud" will give each Xbone the power of three additional Xbones to give you an experience that the cloudless lolsony PS4 can't provide.
Latency aside. Bandwidth aside. Bullshit PR game-enhancing promises that are blatant lies aside.
How is Microsoft going to handle, let alone afford, the "One BILLION console!!" market they are trying to grasp in their iron mailed gauntlet?
Let's say, just for shits and giggles, that the entire 300,000 server infrastructure that makes up Azure is used solely for Xbone cloud whatever. This excludes every enterprise application, every private user, lab, everything (which I would assume is a much larger user than what we would believe, as well as a much higher paying customer) that is not Xbone related.
Even if Microsoft launches their Xbone and only sells 500,000 systems (which is a horribly low number for what they're trying to accomplish), each of those 300,000 servers would have to be MASSIVELY powerful PCs to make up for the "3 Xbones in the cloud to work for your Xbone" claim.
Would it be a conservative estimate that each of those servers would have to contain at least $3,000 worth of enterprise-grade parts to even come close to meeting the projection of supplying the initial launch of 500,000? Is it even possible to build each of those 300,000 servers with enough high-powered parts to equal 1.5 million Xbones? We're talking that ideally, each server would have to be as powerful as 5 Xbones to meet the "3 Xbones for every sold Xbone" on 500,000 Xbone sales .
Even if I were to build a personal computer that had the processing power of 5 Xbones, it would cost me over $3,000, and that's with PC parts, which are a fraction of the price of enterprise-grade parts. I have no idea if Microsoft's ability to purchase huge orders of Dell servers counterbalance the enterprise level pricing, so this is just a completely uninformed assumption on my part. Please take this with a Dead Sea sized grain of salt.
Each of those shipping crate sized server houses hold anywhere from 1,800-2,500 servers. At $3,000 each, you're looking at $5.4 million-$7.5 million per server house.
For 300,000 servers, you're looking at a total of $900 million.
Even though they can add on additional server houses as needed, each 2,500 maxed out server house would cost an additional $7.5 million dollars to add onto the stack.
All of this is supposed to be subsidized with XBL? Can Dell (or any other company) pump out thousands of Azure blades each month to meet demand?
This makes no sense whatsoever. To meet what Microsoft's boasts of the power of the cloud, each of those 300,000 server blades would have to be Cray level supercomputers. Hell of a deal for my completely hypothetical assumption of $3,000 each.
Again, this doesn't include every company, every lab, every single other customer that Azure was originally built for, all of which use gargantuan levels of space and computing power for their own purposes. These 300,000 servers are exclusively for Xbone use only.
What happens if they sell a million Xbones? 10 million? 100 million? All of these numbers are horrendously lower than what Microsoft is projecting if they're planning on making money on all of their world-conquering plans, and yet, it seems like the cloud isn't big enough to accommodate even 1/100th of the Xbones Microsoft wants up and running if they're going to make money.
You can't make the argument that "not all Xbone players will be on simultaneously, so they don't need to have the servers available at all times" because, yeah, all Xbones will always be connected. I highly doubt that Azure will be programmed to disconnect a Forbes 500 company's VM to make space for your CoD multiplayer game, nor will they stop some cancer lab's continually running gene-crunching program so you can be too lazy to drive your own video game car.
There's way too many things that make absolutely no sense whatsoever. There's too many "what-ifs", and Microsoft has been going out of their way to not say anything outside alternating between shooting themselves in the foot and espousing dreams made of unicorn love and ice cream sex.
Seriously, in my own personal opinion, "the cloud" will have to do very, very little, if anything at all, to accommodate the number of Xbones Microsoft wants connected to the cloud. It would be absolutely ridiculous to assume that these server blades are capable of computing on a level to "enhance" the game through sheer computing while simultaneously accommodating all the needs of the enterprise-level accounts who pay shitloads of money to utilize Azure.
I'm sure there will be cloud saves, leaderboards, the occasional MMO-type server-side programs, and of course the authentication pinging, but for there to be serious amounts of number crunching for a game seems highly doubtful to me, especially when big, new games come out and there's a million people trying to create a VM for their deathmatch map.
Am I wrong? Am I completely off? Do I not have a clue what computers and clouds are?
As someone else mentioned, the biggest advantage is in using a large number of CPUs to do some calculation very fast and then releasing all those assets so that the total cost is low.
IMO the high power of the XBO makes it harder to leverage the cloud in this way.
(Please do not gift. My game bank is already full.)
Governments wants cloud computing sooo bad so they can snoop even more of our data. Corporations want cloud computing so bad so they can buy/sell/trade your data with other entities (including government)
I believe the GNU creator, that cloud computing is a trojan horse to get your data out of your control and away from your hands. Cloud computing is a way for companies to monetize services that should be free that you take for granted on your own devices.
i don't think gaming will ever really benefit from cloud computing... besides just the standard fair we have seen over the last 10 years, like centralized servers serving up multiplayer game content (WoW, FPS)
No, all of this is subsidized with Office. Putting XBL on Azure means that their hardware is still being heavily utilized outside office hours. They may be expanding Azure, but the idea that they're doing it solely for XBL is purest spin.
Hmmmm, I stumbled onto this thing. Seems like a Azure server that might be up to the power of the XBO cost about $.40 per hour. If, say, MS was making 3 of those available for each XBO per their PR and you played a 8 hour game that would be Something like $10 in cloud costs for your playthrough.
Now I don't know if that should be discounted because of non-peak rates. Then again, 50 million XBO might turn work hours into the non-peak times for azure.
Just doing that rough calculation makes me wonder if MS can realistically provide significant computing power with Azure if the XBO allows used games. The could would turn the situation of not getting any money on used games to one where MS (or publishers if they are charged) lose money on used games.
Of course none of this means dedicated servers or cloud saves or whatever would be uneconomic. I'm just talking about significantly increasing the computing power of the XBO.
(Please do not gift. My game bank is already full.)
the actual cost is much much lower than what they charge outside people to rent their machines.
We aren't getting Microsoft's internal cost metrics because why would we?
Lets just say that the 60/yr people pay for live gold will likely break even the expense of running Azure for them, if not net a small profit, as it value adds to their service and convinces people to purchase more games, especially through the digital distribution service if they can really nail that.
Let's play Mario Kart or something...
(Please do not gift. My game bank is already full.)
From a review.
Might be cool, no?
Problem is, it's shutting down at the end of this year.
MS is dumb for even letting something like this happen as they try to launch a new console with "infinite power of the cloud" as one of the talking points.
(Please do not gift. My game bank is already full.)
But cloud games shouldn't ever need to be shut down. Whenever the cloud is needed, a VM will quickly be generated to host that game briefly, now and for the foreseeable future, right?
(It's about 1:17 in)
The interesting thing for this thread is that he expresses total panic at the idea that the code for their games actually be run in parallel outside of some special circumstances (basically completely trivial cases, something like there are X particles, do Y to each but not letting the particles interact). In fact he suggests an complete different programming paradigm would be needed. He mused about how perhaps he should have made up QuakeScheme instead of QuakeC then that might have made things easier because basically the little mistakes that C or C++ let you make would start freezing the game at random times. He's actually talking about using multiple cores more efficiently but I think it's interesting for cloud computing in that cloud computing has basically the same problem but it's even worse because it's less predictable. Anyway the extremity of his solution might convey the level of difficulty in writing programs that both run in parallel and interact with the user.
On the other hand he is a big proponent of the idea that having more computing power available will make things simpler for both programmers (which is obvious) and content creators (not as obvious but basically less tricks are needed when you are rendering things more realistically). And that of course is part of the long term benefit of cloud computing as we discusses a while back.
(Please do not gift. My game bank is already full.)
Your concerns are valid, but at the same time you're assuming that Azure is something that was created specifically for XBone. It already exists and processes data for close to a billion users daily.
You're also making the assumption that every user around the world is using all the bandwidth at the same time.
I don't believe the claim that the cloud will triple the power of an Xbone. But I'm pretty confident that Asure has the infrastructure to handle the real world demands of what Microsoft can expect.
Edit - woah oooold draft slipped in there, deleted now
Steam: adamjnet
Sounds about right, maybe a touch on the paranoid side, but basically it's not really about efficiency. If it's so expensive to host server farms and bloo bloo bloo look at the size of this cross we must bear, then why are they so eager to make video games (like simcity) that could have simply been single play local games into online monstrosities? Not out of generosity or desire to deliver more features.