"Trailers and feature-length movies simply have a much higher budget per second than what the full game can afford," says Filmic Worlds boss John Hable.
The idea is tempting but simply offloading assets to cloud won’t work, says Filmic World's boss and Ex-Naughty Dog developer John Hable.
Replies(12)
I can’t comment on any specific applications
In other words I can't really comment what going on at MS because I have no ideal what they are doing. -___-
Well, the last demo they showed was cloud based physics calculations. They never said anything about graphics processing in the cloud. But they have demonstrated other uses like cloud based software updating and cloud based physics.
But indeed, the issue is that Microsoft hasn't actually shown anything new or compelling yet. This is part of the problem. They made a lot of comments at the beginning of the gen about the potential but it hasn't come to fruition yet.
http://www.pcgamer.com/nvid...
He must have forgot nvidia was all so messing with cloud computing with lighting effects in games. So I call B.S on his statement.
@vega275 That is not graphics though its just doing the calculations then sending the data back to be rendered.
“for computing indirect lighting in the Cloud to support real-time rendering for interactive 3D applications on a user's local device.”
says so right there
And he would know about everything that's going on behind the scenes at MS. I mean he does work for Naughty... Oh wait. He wouldn't know a d@mn thing would he? lol
Replies(3)
This is not a specific comment on MS... It's to do with offloading graphical computations to remote servers.
Did you even read and think about the article or did you just see the thumbnail and go into "console war mode"? Pretty low of the submitter to do that but I would expect almost everyone to see past this media manipulation by now. RTFA.
Of course there are limitations in trying to perform low-latency graphical computations over a network - this is all that has been stated and is common sense.
It will be cool to see what does emerge from cloud computing in terms of gaming. Maybe some AI can finally make big leaps, esp. as the generation unfolds and there are more servers and power.
Tbh it is hard to think of something that isn't required almost immediately in gaming though - I don't know what the average would be but something like a >100ms round trip time seems reasonable... and is an awfully long time. But perhaps there are tricks and techniques to overcome this. Maybe some things can be calculated in advance for the next few frames.
It doesn't make sense to even talk about offloading graphical assets to the cloud since that was never the goal of cloud compute in the first place. It was meant to take care of some of the CPU tasks (most very small) in order to give the CPU more room to take on other things. The cloud is not rendering graphics, the console is still doing that. All it's doing is taking bits and pieces of data that the console needs for things like AI or physics, and sending it to the console. Not rendering it in the cloud.
The paper that Nvidia provided with their demonstration has a lot of great information on what is going on, as well as the requirements to make it work.
http://graphics.cs.williams...
It's a pretty technical read, but good none the less.
Nvidia claims that the standard ping time in latency is sufficient in most cases, however, the actual real world results would be highly dependent on a lot of things. Certain applications work better with lower ping.
This of course only took into account 3 different types of lighting systems, but they are fairly common approaches in today's games, but often don't happen on the same scale as it'd be resource prohibitive.
This is of course only one implementation of this kind of tech. There are others out there, and I'm sure there are some that have yet to be revealed.
@Lennox
That's actually true. However, Wardell, in his infinite wisdom, decided to discuss offloading the lighting, which many in the media took as MS actually saying it. MS has mentioned this procedure, but never specifically mentioned it in regards to X1. Much of the expectations from some people on the tech comes from misrepresentation from a 3rd party who was only talking theoretical possibilities and not actual intented application.
Cloud compute to MS was 3X the resources, however resources does not equate to 3X the power. For example, 3X the resources in a physics engine means that you can calculate 3X the number of physics calculations per interval. 3X the power means that you could likely process 300X the number of physics calculations per interval, or process much more complex physics calculations than what would be necessary for a game.
That being said, whether or not games need to calculate 3X times the amount of physics per game is questionable. I imagine there are times when it could be nice, but not sure the overall practicality of such a feat as there does need to be an object for everything physics calculation.
AI makes a lot more sense, as it can be quite complex, but the results, and variables to derive those results are typically very small and allow for quick and easy transport through standard latency scenarios. Any lag introduced would likely be imperceptible to the end user, unless there was a huge spike or disconnect, and there would likely be backup routines should the data not arrive in time.
It does make sense to talk about offloading graphical assets to the cloud if you're asked the question in an interview though lol. And it's also an interesting topic outwith any narrow console war you may or may not be involved in.
The goal of cloud computing (in particular console gaming) is to provide a better experience to the user by working around the limitations of the hardware in whatever way is feasible.
No doubt work is being done on distributed graphics right now and if it is a feasible option, you can bet it will be tried at some point.
Asking the question to a developer (ex or otherwise) for their thoughts on this subject seems okay to me.
Lennoxb63 says, "It doesn't make sense to even talk about offloading graphical assets to the cloud since that was never the goal of cloud compute in the first place."
Ars Technica interviews Matt Booty.
http://arstechnica.com/gami...
I'm sorry, what was that?
I like how old developers from Sony are commenting on the new things MS is trying to do. lol
I don't see a direct reference to, MS... You just made the assumption that the cloud tech is exclusive to them.
That being said, I think it's actually Sony who are pushing it forward at this point with game streaming. I've seen very little other than a couple demos touting one thing or another from the super bs pr team over at MS.
Game streaming is as different from cloud processing as playing games online.
MS has built the XB1 architecture around this function. Each of the 4 multipurpose DME's can do data movement at 26Gb/sec to/from any 2 locations at no cost to CPU, GPU, or memory bandwidth. My understanding is that server systems use these types of accelerators to communicate between nodes within the server bank. MS will show Crackdown at E3, so they will talk more about the cloud's usefulness in gaming then.
Read the article and to sum it all up it all depends on internet speed and reliability.
Cloud graphic and physics improvement is still too early because of the slow and unreliable internet speed that 80% of gamers have.
MS is onto something but its way ahead of its time.
This, exactly. Microsoft is right, but it's not practical yet, but when it becomes practical it will be amazing
Eighty is way overblown. The Washington D.C. Baltimore region and northern Virginia have over 10-13 million people. The average internet speed here is 30mbs easy. I would bet most of the major cities are similar. I would bet a third of this country has around that speed or higher. The problem is low population states like North and South Dakota and Montana bring the average down. And we have too many states like that. But more than half of us live in the major cities.
On topic I think this guy is too busy with his new business to be up on what Microsoft is doing. Some of his comments make that clear.
He mentioned server cost.
It's like he doesn't know what Azure is. So the answer to who will pay for all those servers is the companies that use Azure. Microsoft offers them to developers for free. He has a business, he knows business' pay a premium for every service they use.
Bandwidth.
Microsoft and Duke university have already cut back bandwidth need by 83%. Plus if he looks up some of the information on Project Orleans a big part of that is instantaneously hydration and dehydration of information to reduce bandwidth needs.
Wi-Fi?
Not even sure that is a real problem. If your Wi-Fi sucks buy a long cat-5.
Server goes down.
Again Azure has protocols in place to switch anything running instantly over to another server. He has to remember Azure is being sold as a business tool with Quality of Service guarantees. They want to use this to create mobile disaster infrastructure that can quickly switch over to a new host if needed. Look up agents in Project Orleans.
Latency is an issue.
Some of the other things will help address that. But the predictive technology Microsoft has been working on may play a part in that. As well as lost packets.
The reality is it may be really, really hard. But there are people thinking outside the box working on it. One of the things they were working on was the console handling everything in a certain area around the player and the cloud handling the things further away. Plus I think people should remember some of the rumors we've heard over the years. It could be a matter of the cloud just adding more details to things the console draws.
I think people need to wait a couple of more weeks.
I think the round-trip time is the big problem though, not bandwidth or reliability. And unless servers are going to be in your country and close to your location you might not be able to benefit from cloud computing.
It's a really interesting problem to overcome. So many variables that will affect everyone's experience differently.
50ms RTT vs 250ms RTT.
Internet cutting out.
Internet traffic at peak times.
I wouldn't like to be the guys programming that... actually that's a lie it would be awesome but very very difficult!
It's always funny seeing people talk about RTT, but never bring up how MMO's receive data, update state and game logic, apply data and then return said packet.
Why, if a Cloud Compute approach is held back by RTT, would any MMO be playable if it requires a similar approach for data?
That's a good point... also any online game in general must do the same.
I'm not an expert on this at all but I think there are a lot of tricks to achieve this and that's why the experience is not always optimal and varies per game.
For example: Client-Side Prediction and Server Reconciliation discussion here ( http://www.gabrielgambetta....
This kind of workaround to lag may not be possible because you would need the result (at least for graphical computations) immediately. But maybe things like AI could lag a few frames behind and be better than local AI.
It could be a great thing for some people, just like game streaming could be, but in my experience with my internet connection the lag sometimes creates noticeable problems.
In terms of physics, beyond a feasibility standpoint, the developers still have to want to implement these things on such a grand scale. Implementing all these little extras takes time and resources that honestly could often be spent better elsewhere. What's the point in having a billion pieces of a destructible window, when one million will be sufficient?
Over a great period of time, obviously things will become available which make this kind of stuff more feasible on a development level, but there comes a point of diminishing returns. No matter how much a computer system may be able to do something, that something still has to be implemented at some point, and that takes time and money. One of the basic tenants of AAA game design is to make the complex out of the simple, because the simple is cheaper and more flexible across multiple implementations.
In graphics there is a term called "Level of Detail", or LOD. The premise behind this is that objects that are very close to the user's view have a higher level of detail applied to them, whereas things that are very far away have very little detail applied to them. The same is true with physics calculations. How many objects can a user reasonably have within their immediate view that requires such vast amounts of physics processing to make move? Again, diminishing returns for what amounts to lots of work on the development side.
On top of all this, for graphics rendering, GPU's are becoming more powerful faster than the internet infrastructure is becoming faster and more ubiquitous, so over time, the idea of rendering on the cloud may actually hinder the designed almost routine abilities of a graphics processor. The idea of remote rendering makes sense on certain types of devices, say mobile due to the issues with heat in a very compact device, as one day, those devices are not going to be able to go any further based on today's technology. Because of this, Moore's law is actually coming to an end at twice the speed of the average PC component. However, should a device have the ability to have a rather recent GPU, then chances are that it's abilities are going to far outpace the abilities of what the extra cloud rendering could provide.
John Hable on how DX12 will impact consoles and PC.
Why would a random ex-naughty dog dev who isn't working with DX12 on XB1 or PC or gaming at all currently (film graphics) be a reliable source to quote from? Seems pointless.
Not to mention both consoles are strongly CPU bound as their core speeds are terrible. Not saying he is wrong or right, but someone who is actually using the software would make much more sense to quote, these comments from him are essentially guesses.
I feel like gaming bolt decided to interview this guy for no reason other than he worked for Naughtydog and wants to start a flame war.
"The short answer is that newer APIs will make the CPU faster, but will probably not have much effect on the GPU,” Hable said to GamingBolt. “The improved APIs change how long it takes for you to tell the GPU what you want it to do, but they have no effect on how long it takes the GPU to actually execute those instructions. “ - ex naughty dog dev
"
They might be able to push more triangles to the GPU but they are not going to be able to shade them, which defeats the purpose. " cd projekt red
http://www.cinemablend.com/...
Same thing?
An apu is gonna suck no matter what you do to it.
Sorry.
What will dx12 do for a toaster?
What will dx12/Vulkan do for a gtx 960 and an i3/i5? Or an I7 with a r9 390x with HBM?
Now that is a worthy discussion.
You're getting disagrees for being right...the fanboys are strong with this article.
The reason the CPU's are struggling in the new consoles is not because of the overhead...it's because the CPU's are weak.
A $229 CPU from 2008, the i7 920, at stock 2.66ghz let alone the easily achievable on air cooling 4 ghz runs circles around the consoles.
That is why the CPU's are struggling...it's because they are weak.
8 cores doesn't mean much, when each is pretty weak.
Or else some people might think you can take one of the new octacore chips in a phone and think it's powerful.
I love my PS4, but a $329 GPU added to a near 7 year old PC will run circles around it.
I don't see why people still are in denial, we've had these discussions for nearly a year that PC's will get a much bigger boost then any console, because PC's have the overhead wasting the power they contain.
Why people don't think Naughty Dog or even a layman can't know this and needs further proof is the height of ignorance.
So when developers who never worked with DX12 say DX12 will do nothing for the GPU, their speculation is magically right? Brad Wardell, on the other hand, has worked with DX12 and he is repeatedly discredited by people here. Love the double standard.
We'll see at E3 who was right when DX12 gets its XB1 debut.
Well I'm pretty sure both AMD and Microsoft were well aware of the GPU bottleneck with the old DDR3 GPU prior to them building the custom chips for the XB1. I'm pretty sure M$ prepared the XB1 to handle that bottleneck that Mr. Hable discussed. The XB1 was never built to run the low level DX11 API, which is pretty obvious by the earlier games. It was build with DX12/Win10 in might. The system launched some 18 months early. So we'll see what happens after the key is inserted into the system.
LOL Quoting CinemaBlend as a source, pathetic that site is garbage and is run by fanboys
And I'm supposed to believe some random guy that "USE" to work for Naughty Dog? Nah, Pass! Microsoft isn't going to "cheerlead" directx12 unless it actually does help out the Xbox in a noticeable way. If it didn't it would come back to haunt them worse than E3 2013 and Phil Spencer isn't that stupid.
You know whats pointless? Microsoft's cheerleaders talking about how dx12 will change the world and up the graphics and framerate... even though dx12 isnt released yet. Neither is Windows 10 or any games withdx12. If anybody can put in their 2 cents in so can he. Deal with it.
I agree, those people are also stupid. Are you suggesting fanboys in a comment section somehow make an article based on an interview with someone who has no experience with the software they are talking about is validated by such?
If so that is just as stupid. In fact, you claiming it will do nothing is also equally stupid as you have no clue, and it has been proven that there will be improvements. (Just how much is the question) The reality is, the software is neither released, completed or implemented yet. How about we wait and see instead of pumping more into it. People are going to believe what they believe until shown otherwise so these articles are pointless unless given by someone using it.
Funny that you think your assumption is better than everyone elses... to the point of chastising those who have different assumptions, and excusing those that don't.
@Tsubasa
That would be true......if there weren't dozens of benchmarks already shown to the world...
...but there are.
And we can already see a massive improvement from DX11 to DX12.
There will be less cheer leaders if there is less haters.
*logic*
MS fanboys arent just console gamers which mean there the ones getting the best out of DX12.
@Nicksetzer
And yet you have no idea, what experience he has with DX12 or what he knows about it. You do know that a lot of these guys all know each other right? And that they all talk right? Im pretty sure that by now everybody who is anybody in the know, knows a whole lot about DX12. Its not like a low level API is even some new mystical thing. Its been done in consoles forever now. So let's not fall back on that default fanboy argument of "how would he know if he hasn't used it yet"? How you dont know he hasnt?
@Out
"And we can already see a massive improvement from DX11 to DX12"
On the PC! repeat, PC, not X1. You are making assumtpions. We don't know what it going to do for the X1.
well of course they did ,i'm just surprise the click bait went the other way this time its normally asking every indie dev why the xb1 sucks lol he really has no idea what tools are in dx12 for xb1 .
lol, both are strongly cpu bond, lol. The weakest part of both is the CPU, average mobile cpu...
Well... not "average mobile CPU". More like the latest smartphone.
The Galaxy S6 (which was released recently) has both a Quad-core 1.5 GHz and a Quad-core 2.1 GHz. The PS4 and Xbox One Octa-core CPUs are respectively 1.6 and 1.75GHz per core.
Sure does look comparable, though.
He's a high level developer who worked at one of the most difficult game companies to get hired at. He sure as hell knows what he's talking about, whether or not he's currently developing something on it or not.
People seem to be fine taking the word of MS employees who aren't even developers regarding DX12. I'd hold the opinion of a talented, unbiased developer a little bit higher.
He maybe a high level developer but he still hasn't had anything to do with dx12 and why does it matter if those ms employees aren't devs as at least they actually had hands on with dx12 yet your saying that his opinion on something he has no first hand knowledge of is more valid than those who have
Many take Wardell's word for everything, despite him clearly stating that he doesn't know the X1 well enough to say for certain. I can respect Wardell's comments because he does at least have the knowledge to reason out what is likely to be the case, and I also realize that many of the things he says simply get attributed to the X1 despite most of the time he's only talking about PC.
This guy could probably get picked up at any MS studio if he wanted to, and go in without more than a couple days to get up to speed on DX12 specific syntax. People really don't realize how talented game developers have to be to get jobs at studios like Naughty Dog. It's not like he was some sort of intern who worked there for 3 months working on linking the menus to different parts of the game.
I've done both console programming and DX12 programming(for PC) and I can tell you there isn't a major difference in how operations are handled between the two. Syntax is different, implementation is different, but DX12 operates pretty much the same way consoles do.
It's funny though. This is a great break down of a major difference in DX12, and it's a great thing, but some people are more concerned with discrediting the statements without the knowledge or the research to refute it with something factual.
@rain c++ and visual basic are the primary syntax languages for dx3d of any kind, very few changes in that. Weird you claim to be some god-like programmer but don't know that...
Not to mention crytek, unreal and square have all had tech demos showing there is an inpactful change. So should people believe you (the random self proclaimed pro) or the people who actually presented something with the software?
http://m.windowscentral.com...
http://wccftech.com/king-wu...
So if you want to believe it does nothing, enjoy your misbelief. The only question is the effect it will have on XB1.
I'm with Nick! It would be dumb to take serious the words of those (with even huge reputations) that have absolutely no experience dealing with dx12 over those that have some.
By learning new sytax I meant the new functions that exist within DX12. For console programming, they're either going to use C, C++, or the assembly API. There are also some game engine scripts that they will likely use, and many many 3rd party tools which will get licensed to make things work. Visual basic won't be used because it works off a framework which isn't suitable for AAA games, but can be used for simpler games.
When I say there isn't a major difference, what I mean is that overall, the differences are on levels that aren't actually programmed in individual games. To the average developer, they're just going to use the engine, and then provide special functions through the low level API if necessary. It's EXACTLY the way console programming is done now. Not much will change. Sorry for being unclear.
I never claimed to be some god-like programmer. In fact, that's my point. It doesn't even take a genius programmer to go from one to the other if you know the basics of one of them. No people shouldn't listen to me, but they should at least verify or research what I say to determine for themselves if what I say has merit. I don't often dismiss other people's comments without at least trying to verify if they may have some merit.
Did crytek, Unreal, and SE show off anything for the X1? because that's really what this discussion is about. This guy is discredited for his work at ND, yet you point to all those developers who haven't made any DX12 games for X1? Seems legit.
On PC I have said many many times the differences in how it operates are substantial. I can attest to this based on my own work, and I am very impressed at what it can do, and I'm a little miffed that PC's have been gimped for so long due to this kind of stuff not being available years ago without 3rd party tools.
And that's what I'm saying. DX12 brings to PC exactly what has been on consoles for decades. It's a touch more higher level, but it has an extremely efficient low level API, just like consoles.
Let me know if you want to misread and misrepresent my comment to discredit me some more. I'll be happy to respond.
If you wish to continue on with your eyes closed, and fingers in your ears going lalalala, while ignoring anyone with a comment contrary to your own, and can't bother to provide me with any kind of response that actually does address anything I say with factual information then please just put me on ignore. Let my comments be used by those who want to take the effort to learn more, and not be like yourself where you throw out a few computing terms hoping that you seem knowledgeable enough to discuss the topic properly.
I don't care what your perceptions of what it will do are, but I do care when you make it out to be something that it's not. You set an expectation for others which MS can not possibly match. At least do me the common courtesy to respond with actual facts pertinent to my comment, and not with more exaggeration and PR nonsense which only validates what you already want to believe.
What??? MS built DX12, but some guy who worked for Sony is more credible? That's just stupid.
I'll tell you what. I've got some gold to sell you... yeah, I know the Periodic Table says it's lead but they don't know what they're talking about. Trust me.
Come on now. It depends, but for the most part those MS employees actually work at Microsoft and are communicating with and/or have had experience with Dx12 or those working on Dx12, unlike the ex Naughty Dog dev. I mean geez the difference is that simple.
Man, I wonder how you all would react if some ex Halo Dev started talking about Nintendos new API or a mario dev started talking about Vulkan. Tune would change here.
And there is a reason no one else is talking about this besides Gamingbolt lol.
Why indeed? Probably because gamingbolt asked him. He's not wrong with what he says though. It's not like you need to know each specific API in and out to be able to work with them or understand what they'll be doing. Good game developers don't live in a bubble, and they all have access to forums or documentation or personal contacts where they talk about this stuff, and if they're good they keep up on everything that may be available to them.
I knew nothing about DX12 when I started working with it, there wasn't even documentation available to me until after I started work, but when I had it, I was able to see what it does different from both OpenGL and DX11, and adapt my algorithms to implement what I needed to in DX12. Since that time, MS has provided a lot of that information more publicly to developers if you know where to look. It doesn't make their PR rounds as much though.
Also, consoles are not CPU bound. They haven't been for a while now. PS3 was designed to be GPU bound with offloading onto the CPU/SPE's when needed, and the 360 had a similar setup but did have some overhead with it's memory controller in the cache, nothing major though. PS4 is most definately supposed to be GPU bound as stated by Mark Cerny himself, and I'd imagine the X1 is supposed to be as well, as it's the best way to get the most performance out of these machines.
Otherwise, I didn't notice this was a gamingbolt article before clicking on it, but it is a pretty good break down and simple explanation of a basic difference between current API's and the new ones coming in DX12 and Vulkan. His bunny example pretty much nails it, but he doesn't really explain what overdraw is to help understand why it's not optimal.
This quote is pretty much what a lot of people have been saying for a while.
. “The improved APIs change how long it takes for you to tell the GPU what you want it to do, but they have no effect on how long it takes the GPU to actually execute those instructions. “
In other words, you can't make hardware do more than it's designed to do.
Even Wardell has said something similar.
With all due respect, I'm pretty sure that guys knows a little more about programming than you do.
let me explain what he was trying to say. and im a computer scientist, so please consider that.
imagine you're trying to send a set of order to your employee. you have two options, you can do it via a middleman: asking someone to send the order to the employee, or to send it directly to your employee. naturally the second option is way faster because.. well it's just naturally faster.
programming is a set of order, employee is your hardware, and those middle men is any middleware, such as directx 12. So what directx 12 trying to do is to fasten the middle men, but it got nothing to do to fasten the employee. Does it make the game to be rendered better? of course! but does it make the hardware better? no!
So to sum up, direct x 12 WILL make xbox one's game better, but ps4 is already a better hardware, so it's impossible for xbone to top ps4.
and considering game dev (especially the triple A games) is much closer to the hardware (which minimize the role of middleware), the effect would be very minimum in console dev.
Hmm, I find that hard to believe. Especially considering "computer scientist" is the most outlandish job position description I have ever heard. It would be like calling the people at nasa spaceship guys.
That all said, you contradict yourself regardless. Directx 12 is a kernal and it does transmit data between the hardware and an application, but bettering that process can actually benefit the quality of a game. Will it make the physocal hardware more capable? No. It does however have the same result as it makes the hardware work more efficiently.
A better example would be a helicopters propellers. If you make the propellers directly horizontal they have little effect, but if you skew them slightly they allow for much more lift. Same wings, same engine, same weight.
Proof of this would be to use a simlar piece of hardware running old software and one running new. (Let's say dx9 vs dx12) there would be an absolutely massive difference, despite the hardware being The same.
Point is, we know DX12 is a massive improvement, we just don't know howuch it expands on the current toolset for XB1. From what we have heard the toolset for XB1 is a bit under par. So this could be a two fold upgrade. Stronger API with a more reliable set of tools. Not to mention if XB1 uses DX12 properly it could allow devs to make engines that will accomadate it more readily. Would be like last gen, PS3 better hardware, but xb360 was easier to create for. I think that is the goalS is shooting for. That said, this gen, the PS4 is streamlined enough AND powerful enough that I doubt any gains will overpower it.
It would appear nick, that you want to discredit anyone who thinks differently than you.
While I can't claim ng's credentials, Computer Scientist is indeed an actual thing. It is more on the theoretical side of computing, as opposed to an engineer who works on the hardware or software. A CS would have to have enough knowledge of computer engineering, but the reverse isn't always true.
Here's a list of jobs looking for computer scientist positions.
http://www.indeed.com/q-Com...
Everything he said is true though, whereas your comment is just you throwing out some computing terms again trying to look smart. So lets look at what you say.
"Directx 12 is a kernal and it does transmit data between the hardware and an application,"
No, DirectX is an API. An API is an application program interface. it is simply a set of protocals and routines which facilitate the building of software applications. There are thousands of API's which are used every day, and their main purpose is to dictate how the software interacts with the OS(emphasis on OS).
A kernel is a program which manages I/O request from software, and translates those request into instructions for the hardware.
A kernal interacts between the hardware and the OS, whereas an API interacts between the user(software) and the kernel. The idea of low level access bypasses the kernel to a degree, but in a PC that's not ideal as it can cause serious security issues, which is why you need a low level API because opening up a system fully to low level access can cause all sorts of unwanted side effects. I don't even think consoles allow for complete bypassing of the OS kernel, unless it's through the hyper-visor. Outside of closed systems, I can't think of any API that directly interacts with the hardware. Almost everything is done through a level of abstraction, and DX is no different. The whole idea behind DX was to remove the hardware from the equation to make games more compatible and easier to program for PC.
" but bettering that process can actually benefit the quality of a game"
Yes. And DX12 does that. that's what the article says, that's what I've said, that's what ng said. It can benefit the quality of the game, and it will, but the effect is that the hardware can get more data faster, or more efficiently, or spend less time waiting for data to perform it's task.
That isn't the same effect, because better hardware can do more, whereas the current hardware can do what it's designed to do. The efficiency is that it's not waiting on data, or that the data can be output without waiting, or the data can be supplied in a more acceptable manner for processing, but the data itself is still processed based on the rendering algorithm used. I know it's a nuance, and you aren't exactly wrong, but you are making assumptions about the gains.
"From what we have heard the toolset for XB1 is a bit under par"
What we've also learned is that an implementation of DX12 is already installed into the X1 API. It's not the full Dx12 because it's still not finalized, but the core is implemented from what I understand, and the current low level API probably isn't going to change either way, as the current LL API was already built for the X1, never existed in DX11. Same as last gen, the X1 has a custom version of DX.
@nick wow... Are you being serious... A computer scientist is technically anyone with the job position titled that (research based I'd assume) and/or anyone with a degree/studying for a degree which here in the UK anyway is known as "Computer Science" I know several people studying it currently as I myself study physics and therefore know many people at uni studying other sciences because... Well your just more likely to meet them. You seem to like discrediting others comments without any real evidence so I'm going to assume you have no background in any form of science because if you did you'd know. Evidence is key, without some tour credibility is zero. And both the guys above seem to have evidence for their comments and potentially experience in the field. Same goes for the guy being interviewed.
Computer scientist is not a posistion nor a degree. It is a description of such. Anyone who has a degree or position would know that. Generally all it means is a major in mathematics or a computer related field. Really simple to understand. Just like saying you work on broadway. There are lots of jobs in broadway not just one.
http://en.m.wikipedia.org/w...
The xbox one "beta tested in the future" was specifically made with DX12 in mind. The Xbox One is a windows 10 device. DX12 will not only make use of all the cores (Currently only one core being used with dx11) but it will also make much better use of its super fast ESRAM. Tiled resources, etc...The main reason for the difficulty for some 3rd party games not reaching full 1080p. Moving to DX12 is huge in this regard and it's something many conveniently leave out when discussing the true benefits of games developed using DX12 on the console.
Another huge benefit not talked about much, but expect MS to really do a deep dive into this at E3. But the sheer ease for developers when using DX12 on PC and literally with a push of a button, your game is now scaled & optimized for ALL windows 10 devices. Devs will save money, time, etc...And consumers/gamers when they buy a game on ONE windows 10 device, you own it for ALL your windows 10 devices. That to me is the game changer.
Lastly the fact that if a game is built using DX12 & windows 10, cross play will be simple & easy. For years I've always dreamed of a day when PC gamers could play with console gamers. Well after July that dream will be a reality.
Dx12 & windows 10 is definitely a game changer for xbox one. The xbox one will finally have software for developers that the xbox was built for from the beginning. It's amazing to me how with the low level dx11, the xbox one has been able to keep up with the Ps4 and in some cases do better. Fanboys say that xbox one can't do 1080p. Strange, because I have over 10 xbox one games that are native 1080p.
Dx12 will open up all kind of doors for the xbox one. Soon enough, gamers will see.
he probably knows a thing or two about PC programming numbnuts. what do you think the PS4 and xbox1 basically are?
Oh gamingblart. Playing both sides off against each other again. Classy stuff.
Strange that anyone with even a modicum of API knowledge has been saying what's posted here (and more) for months. The whole while you've been posting BS about massive gains to Xbone, pc and even mobile.
Serious question? Do you ever feel bad for the hit whore tactics you employ? All they do is cause division and angst.
What was that old chestnut about the ends justifying the...something?
Edit: Lol at the damage control attempt by nic (first comment).
This isn't a Naughty dog dev. It's the guy who runs filmic worlds (a graphics solution based company thats open to pc mostly). He worked for Naughty dog at some point. That (the ND reference) was used to generate hits and cause an argument. Seems you fell for it..'coz...you know. Creating fighting fanboys is how this site does business.
Edit 2: nice edit nic. Keep up the good work. *smfh*
Just think, a pact of script kiddies are going to tear this news outlet apart along with Gamingbolt.
Honestly gaming bolt is a farce of a site. Ppl put way too much stock into the differences of direct x and Open gl, when a new version comes out of course it's better, but in the end it all pretty much evens out.
I think the ex-naughty dog dev is mixing old technological advances with today's new technological advances of gaming & hardware/software.
Thing is, MS took a completely different approach with designing the x1's hardware.That said, They simply went above the standard engineering process to make some of the x1's parts custom built. By doing this, they gain a bit more hardware control over the x1 CPU/GPU, and how it should send n process data, back and forth.
The x1 was design with dx12 in mind. That makes a world of difference. The way I look at it we don't know enough about the x1 hardware to talk about it capabilities.
That's what has me interested. We don't know enough about the hardware, and whatever details we don't have I feel like we will around the time windows 10 & Directx 12 comes out. I'm not saying there will be some magical boost, but like you said the X1 was built with Directx 12 in mind. They planned it out from the beginning and all they talked about seems to go together like pieces in a puzzle.
It's all very interesting imo.
It's hardly custom built just because it has esRAM..
They didnt take a completely different approach whatsoever. It's literally just like the 360. Just stronger, obviously with added functionality.
I don't see anywhere where they went above any standard engineering process. It's a console. Of course, it has more hardware control access, but it's no different than any other console. It sends data back and forth no differently, aside from esRAM, which requires work to get anything out of it.
Yes, it was designed with DX12 in mind.. no. Wait. DX12 was designed with consoles in mind. Not the other way around. DX12 gives PC's the lower level access that consoles already have. That is all.
Nothing is revolutionary about it's design.
Nothing is revolutionary about it's design. Oh really?
The x1 GPU is custom built. Its the only GPU in the world that has dual lane. This tech feature its the first of it kind. There no other GPU like it. Beside, MS wouldn't spend 2 billions bucks on X1's custom built GPU for no reason. Keep in mind, you can't test something that isn't readily available on the market. That said, I can clearly see why MS had to wait until x1's custom built GPU is born before writing DX12. I think had MS used a regular GPU for the x1's design they couldn't start writing dx12 way before the x1's hardware were completed.
http://www.reddit.com/r/xbo...
Another thing, Illogically speaking, what I find obvious about the x1 custom built gpu is that it has dual lanes which means that some of the other x1 hardware component has to also change to sustain data communication between the two lane that the x1 GPU has.
I'm guessing, that why MS had to implemented a move engine to take full advantage of those two data lanes. I could be wrong on that note.
Anyways, my point is the x1 has some hidden mystery about the x1 hardware. So, to make claims on what it can do it plain crazy thinking, especially if you don;t have(know) the x1 full specification.
Jhoward...what you've just written and linked to is complete crap.
The "dual lane" you're waffling on about is for gpgpu compute and refers to ACEs. They're used for parallel computing. They each run 8 queues (lanes). It has 2 ACEs for a total of 16 queues (lanes). Pretty impressive huh?
For reference the PS4 (the R9 280 and up as well) run 8 ACEs for a total of 64 queues (lanes).
One of the reasons MS would have chosen to go for the older/fewer ACEs configuration is to accomodate the eSRAM. Said RAM takes up more than half the transistor count on the die. There is no secret sauce in the hardware.
Now that kinect has been 86'd they may attempt to repurpose the move engines. Given their ultra-limited bandwidth with no direct access to the eSRAM their usefulness will be limited.
All (yes all) the xbone's spec's can be found online. Beyond 3D and other sites have all that for you if you like.
Now did you really want to talk about hardware or were you just grasping at straws? Tell ya what. I'll wait for your reply and we can continue this if you like.
Here's a tip though. Don't link to Mr X style reddits if you want to be taken seriously. ;)
Edit: LOL all the info the link you provided proves my point if you just look at the info given. OMG! You couldn't make his stuff up (but someone did).
@jhoward585
For one.. You are talking about complete speculation from a rough translation. that still hasn't been clarified.
Logically speaking.. You're still talking about "new" and "one of a kind" tech that isn't actually confirmed, yet, anyways. So, I don't see why you are trying to explain how it works.
Again, still unconfirmed. Don't harp about secret X1 sauce like everyone else is. Always have to grasp onto something..
And, just as LostDjinn said..
So. Yeah.
Love your console. Don't lie about it or find excuses though. The excuses are what games and features you like. Not what you hope it to be.
@sinspirit
YOU:For one.. You are talking about complete speculation from a rough translation. that still hasn't been clarified.
speculation?LOL
ME:Fact is, the x1 does have a dual lane custom built GPU. No one knows how it works but MS.
-----
YOU: Logically speaking.. You're still talking about "new" and "one of a kind" tech that isn't actually confirmed, yet, anyways. So, I don't see why you are trying to explain how it works.
ME: I brought it up because in my mind I know any piece of new technology will eventually improve as it passes through the trail n error phase. The x1 GPU is the first of it kind so there going to be lot of test run on the software side of things to refine it performance.
You:Again, still unconfirmed. Don't harp about secret X1 sauce like everyone else is. Always have to grasp onto something..
ME: I was thinking the same. Until I get more info on the x1 hardware specification I won't take in another false rumor as fact.
You: Love your console. Don't lie about it or find excuses though. The excuses are what games and features you like. Not what you hope it to be.
ME: for the record I own a PS4 not an x1. maybe in time I will.
Another thing, I do my best to make sense of everything I read or hear, especially when I gather information on the internet.
Lies is always a Lies. But one thing I do know is most major company won't take the blame for another company's false(misinformation) claim.
With that being said, MS has made some claims in the past that involve AMD as far as the engineering & hardware design of the x1 hardware.
And yes, I'm talking about the secret sauce stuff that's spurring all over the internet.
Fact is, AMD has their repetition to protect ,and MS has theirs. one thing that is for sure, AMD would've ended some of those false claim to protect their image.
So, That my friend, is all I need to draw my conclusion on what is fact, and what is not a fact.
@LostDjinn
Ok, the link I provided on my previous post may not be the greatest information as far what the x1 hardware can actually do.
Honestly though, I don't think anyone can say what the x1 full specification really are because there's just too many contradiction regarding the x1 hardware specification on the internet. One sites say one thing while another site says another.
Truth is, I was more interested in MS's past business decision they made to fund & create the x1's hardware more than the physical hardware(and specification)itself.
Fact is, MS spent well over 2 billion dollars on the x1 GPU technology. That alone say a lot to me. To me, It means MS took a chance which could've either worked to their advantage or not.
Truth is, we really don't know if it was a bad or good business call. We don't yet have all the details until then I will remain completely optimistic.
I think what u meant before about dual lanes is that the x1 gpu is split into two sets of CU with 2 individual gfx command processors. Thats what brad wardell meant when he referred to xb1 having dual lanes and how ps4 and pc dont.
@LostDjinn
If my memory serves me correctly, I think brad wardell were the one who stated that the x1 custom built GPU cost close to 2 billion dollars.
They way he explained it...The deal between MS and AMD to engineer & build the x1 hardware cost MS over 3 billion. Two billion dollars went into the development of the custom GPU for the x1 while the remainder 1 billion dollars went in to the rest of the x1 hardware component & design.
http://www.gamespot.com/art...
http://www.vg247.com/2013/0...
Genuine- If it's command processors I think you'll find they pertain to OS task distribution. With the discrete (system) OS and gaming OS requiring high and low priorty access. It's they only way they'd be efficient. The hypervisor would be simply access the discrete level OS priorty solution.
Whats the point?
Well, think of something like the snap function. The game would be given a high priorty while say the browser would be given a low priorty (as missing your render budget on a web page would be preferable to the game doing it).
Edit: Jhow neither of your links say anything of the sort. A deal between MS and AMD is all that's mentioned. Nothing about gpus. Please make sure you provide proof of your claims in future. Otherwise you'll paint yourself in a bad light
You just completely edited your comment to cover the fact you can't provide a link. Jhoward I'm now have a very different view of you. It's not about the truth. You were simply clutching this whole time.
Thanks for that.
From that thread, as much as I cared to read since it kind of devolved after a bit, all I can gather is that no one seems to know what the dual data lanes are for.
For those that don't know, a data lane is just a controller for data, and in this case, the X1 has two supplying it's CU's in the GPU. I'm not sure the move engines are really important for two data paths, as it may just add more overhead since the data controller has direct access to memory and the CPU. In the case of the X1, the different controllers appear to control a split set of CU's.
Anyhow, I'm not going to speculate for now, because I'd have to do some more research. I understand what's being talked about, but not enough in relation to the X1 to be able to fathom an assumption. I do think the article which reported the leak was a bit presumptuous.
To me, the best thing to take from that thread was a comment from iroboto,
"It's very important to not get stuck into conspiracy style thinking. It's like if MS denies the functionality of the second graphics command processor you take that as the opposite. If they agree with you, you take as truth, and if they say nothing about it that means you also take that as admission that you are correct. In all scenarios you are only agreeing with what aligns to wishful thinking and that severely hampers your abilities to make good sound decisions."
Not saying you're wrong or anything, just that it may be wise to temper your expectations and wait until more information is out before postulating a conclusion on what MS did and didn't do. I'll also readily admit that I should probably follow my own advice sometimes. There are a couple people on that thread which seem really knowledgeable about hardware, and while they're offering possibilities, none of them are saying anything definite.
Otherwise, I wouldn't really call it a revolutionary design. It's a different design, done to achieve some task which is currently unknown. One person speculated that the extra channel is for the media and overlay features of the X1, so the extra data path may simply be a workaround so as not to take away from the actual abilities of the GPU.
I can think of several reasons why a second data controller would be beneficial in gaming, but it's not a feature of DX12 that I've seen. It may be specific to X1 though, in which case I wouldn't have bothered looking too much into it yet.
However, even with two data controllers, unless there is some specific gaming purpose for them, I can't really think of any reason why you would need one on the GPU's given the rather low number of compute units. The data controller that came standard with the GPU should be able to handle it perfectly fine, since graphics are serial in nature. But when it comes to GPU compute, it can actually help tremendously.
Edit@LostDjnn
Appears you went into more detail about it possibly being a multi-tasking thing needed. While that's perfectly reasonable, I still wonder if it's actually necessary to have 2 controllers. Overlay has such low overhead, it seems rather unnecessary to split the CU's and memory controllers. I can't imagine that system features would run off GPU compute.
Rain that's not where I was going with it. Overlay indeed takes FA overhead. The efficiency increase I was referring to pertains to running 2 controllers with a conventional overlay as opposed to running them hardware based priortization. It has nothing to do with compute based packet distribution to the gpu. Simply the simultaneous rendering of assets from two seperate OS's on a hardware level.
Nice to chat with someone who actually just cares about the facts though. If I run outta bubbles just pm me. :)
Ah. Yeah I think I missed a bit of your comment there. Indeed it does make sense to do that as it would require hardware to be dedicated to the actual secondary rendering to prevent slow downs with the game render. Since console developers have the ability to control memory controllers, it would be reasonable to isolate a secondary controller that is managed by the OS to perform it's functions. This leaves everything still available to the developers and prevents unintended conflicts.
I think one example of why this might be beneficial is looking at the PS3. While a game can render in the back ground when you hit the home button, it can also have stuttery frame rate should it keep running. It's mostly obscured so it can be hard to notice, but it is there in some games. I'd imagine on a multi-tasking view, the stuttering would be extremely noticeable.
It's an interesting approach to handle what could be done with a rather inexpensive secondary graphics chip running synchronously with the main GPU.
The thing these developers are missing is that GPU bottleneck he's talking about won't be in the XB1. Because the XB1 was designed specifically for the DX12/Win10 API/OS there are things in the XB1 that will eliminate most of the bottleneck that would normally happen to a next-gen console. I expect a lot of PC games will get transferred to the XB1 with little problems. I'm interested to see how The Witcher 3 gets updated to Win10/DX12. Most of these developer who don't work for M$ haven't a clue of how the XB1 will handle DX12 because they don't know everything about the hardware.
What these APIs do is to lessen the strain on the CPU. So if you have a game that is bottlenecked by it, which basically means the GPU cannot unfold it's full potential, you get some gains with a better APIs. But the GPU stays limited to it's specs. There is nothing you can do about it.
Example: BF4 with Mantle
Slow CPU + Fast GPU ... gave you a crappy experience in DX11, since the GPU was bottlenecked by the limited CPU. Mantle helped there and freed up some capacity on the CPU, so the GPU could better unfold.
Fast CPU + slow GPU ... no gains with Mantle.
Fast CPU + fast GPU ... if there is no CPU bottleneck, you get no gains with Mantle.
Slow CPU + slow GPU ... minimal gains or no gains, if the GPU was allready at it's limit with DX11.
DX12 will basically work the same way. On PC and X1 the GPU will limit it's effect. It really only shines on old CPUs combined with mid- to toprange graphicscards.
Can current and ex Naughty Dog employees stop commenting on DX12? It's none of their concern. I don't see MS constantly talking about PS4's API.
I know right ps4 guyz talk more about xbox 1 then the system they make games for it.
Graphics programmers aren't bound to platforms like you somehow think they're supposed to be.
Being a graphic programmer doesn't make you an expert of every graphics API. A dev that only makes iOS games doesn't necessarily know everything about developing for Android OS.
Naughty Dog devs have never worked with DX12, any Xbox game, or PC game. So they would know little to nothing about it. So asking Naughty Dog current and ex employees is useless.
@Pandamobile
While that is true, at the same time it makes no sense to ask a developer of a totally different studio, with totally different programming methods. They still don't know much about it. Even AMD; the company that worked with them on it, doesn't even know much about it's capabilities. And you suspect someone who hasn't touched DX12 at all knows something? You can't really compare previous versions to DX12 either. As it unlocks low level access in PCs that has been dormant for years. So completely different case this time.
Every single game developer on the planet has made games for PC. Do you really think that graphics programmers jump right out of university or previous jobs onto a PS4 dev kit without learning DirectX and OpenGL?
Seriously?
APIs are transparent to graphics programmers. Just because they've never used DX12, doesn't mean their opinion on it is completely invalid. They know all the shortcomings and pitfalls of graphics architectures, regardless of whether their employed by Sony or not.
@Lennoxb63
"Being a graphic programmer doesn't make you an expert of every graphics API. A dev that only makes iOS games doesn't necessarily know everything about developing for Android OS."
True that but in the case of Naughty Dog believe it or not, they have Xbox One and X360 development kits at their office. If you don't believe me, look online.
@Lennox
I'll agree with you if you can agree that Phil Spencer isn't qualified to talk technical specs on DX12. He's not even a game programmer, he's a technical engineer of hardware more, yet if he said something about DX12, you would take it as gospel.
Otherwise, graphics programmers that understand how an API works, regardless of which one they primarily use, are far beyond the level of the less astute programmers who simply use pre-made functions to draw a screen. If you think programmers at Naughty Dog don't attend classes and go to things like Build, then you are sadly misinformed. Naughty Dog is part of the ICE team which makes the PS API's and SDK's. I dont know if this guy was part of ICE Team, but do you truly believe that these guys are clueless on different API's which are doing exactly what console API's have been doing for almost 30 years now? Any console graphics programmer probably knows this stuff more than any PC graphics programmer, because it's what they already do.
By your own reasoning, there aren't many DX12 developers out there at all, so no one we've heard from are actually qualified to talk about it in regards to consoles except for maybe a few privileged devs who got early access to it for the DX12 X1 games. There aren't many of those out there right now, and most of it is isolated to engine makers for the time being. So who exactly should we listen to? MS? That'd be fine, but they aren't exactly unbiased. So that leaves no one to listen to to relate DX12 to X1.
I'm sure you didn't even bother to read the article.
He's a programmer who worked at an acclaimed studio, he knows his stuff. Gamingbolt asked him questions, he answered.
This article was a step up from their usual 'posting a series of tweets' style of journalism.
Filmic Worlds boss, John Hable also talks about the selection process at Naughty Dog.
they were nt just talented ..they were also not bored (like many lazy others) to work on PS3 which was a beast ... Kojima,FromSoft and Naughty Dog made their own engines specially for PS3 (also Sony helped them a bit too), which maybe was difficult at start but later were benefited alot by that as now they have their own engines perfected and giving some very well made games ...(kojima started with MGS4engine for PS3 and now evolved to Fox engine, Sony gave to FromSoft the phyre engine which evolved to BBorne engine..etc etc
how is the ps3 maxed out with uncharted 2 or TLoU when there is this?.
crysis 3 on console realtime
http://www.gamersyde.com/po...
http://www.gamersyde.com/po...
http://www.gamersyde.com/po...
uncharted 2 pre rendered cutscenes
https://itani15.files.wordp...
http://media1.gameinformer....
yeah I am blind lol...
Third-Party never maxed out the PS3,they had to change a lot about Crysis 3 in terms of attributes to make it even portable.
Crytek uses high end solutions and do not necessarily optimize them
for pretty much anything,Uncharted 2 was developed with the CBE in mind and took full advantage of it's capabilities,well in comparison.
I would love to see a next gen CBE with more bandwidth and a better graphics solution,it's a very underrated chip
Crysis 3 looked so bad when running. The framerate was in the 15-20 range the vast majority of the time and the jaggies were oppressive.
For all of it's effects, Crysis 3 looked worse than Crysis 1 and Crysis 2 on PS3 - Crytek tried to do too much and the horrible performance ruined the experience and graphics.
You can max out a system, but you can also optimize your engine on a maxed out system. You can only get so much power out of a system. Your next steps are to optimize your engine to get more performance. Which i think you're getting the 2 confused.
Replies(2)
It was obvious that it was maxed out with Uncharted 2 as it will be maxed out with Uncharted 4 and Star Wars Battlefront
I messed up but meant the PS4 will be maxed out with Uncharted 4 and uncharted 3 as both games are coming out in the systems third birthday.
A system can be "maxed out" and still show improvement, as he is saying (I'd advise reading beyond the title). Despite what Uncharted 2 did/looked like, God of War III, Heavy Rain, Uncharted 3, and Beyond: Two Souls looked better. Why? Maxing out a machine doesn't mean you can't make improvements in other areas.
The latter portion of the console's life had developers (as Hable put it), "squeezing out" what they could, though he admits, there wasn't really much left. I don't expect Uncharted 4 to be Naughty Dog pushing the PS4 to Uncharted 2's point on the PS3, but their second game? No doubt.
Not really. "maxed out" means the limit has been reached. If you've reached the outer limits of what it can do, you can't go beyond that. If they were able to tweak something here and optimize something there to get another 1fps out of it or improve textures or improve loading times, it means they didn't hit the max with U2.
What he wanted to say was worded incorrectly.
If the PS3's limit had been 100% reached, you wouldn't have seen better games than Uncharted 2. That is not what maxed out means. Its RESOURCES can't be pushed any further, but the engine code can be modified in a way that would allow for optimization (as plenty on the PS3 proved). The Naughty Dog 2.0 engine had reached its limit on the PS3, but there was still enough juice to make games look and run better after Uncharted 2 released.
The engine, to that point, was maxed. The PS3's capabilities were not. Something I see people get confused on all of the time.
You can max out any machine due to crappy code. It's easy to do, any idiot can do it.
A computer is only truly maxed out once the software has been fully optimized, and that takes a long time as someone will keep finding clever solutions to certain problems.
You're not getting it. If you can max out something that's it. You keep bringing up optimization. If it can be optimized, it didn't hit max. Max is the end of the road. Like I said above, if they can optimize to get better performance, it means it was never maxed to begin with. Max is the end. Being able to optimize means it wasn't at max.
PS4's hardware is 28nm and lowest available in next couple years is 14nm and when foundries are enough experienced with it to reduce initial failure rates to an acceptable level and even on the same as current 28nm then expect two times the performance maximum if you want a console at reasonable performance and power envelope in an acceptable form factor for a home console.
Except everyone is willing to dish out 600$ on a home console.
The Order 1886 already almost look CGI.
This just reminded me how excited I am to see Quantic Dream's PS4 game. What they achieved on PS3 was pretty crazy, I could see their PS4 game getting pretty close to CGI.
I'm fine with ps4s hardware...let's develop great games before we sorry about superficial stuff.
Guy is just butt hurt he doesn't work for naughty dog any more
We all know this and it's 2 to 3 generations away