Drilling Down: DX11 And The Multi-Threaded Game Engine
In spite of the fact that multi-threaded programming has been around for decades, mainstream programmers didn't start focusing on parallel programming until multi-core CPUs started coming along. Much general purpose code is straightforward as a single thread; extracting performance via parallel programming can be difficult and isn't always obvious. Even with talented programmers, Amdahl's Law is a bitch: your speed up from parallelization is limited by the percent of code that is necessarily sequential.
Currently, in game development, rendering is one of those "necessarily" sequential tasks. DirectX 10 isn't set up to appropriately handle multiple threads all throwing commands at the GPU. That doesn't mean parallelization of renderers can't happen, but it does limit speed up because costly synchronization techniques or management threads need to be implemented in order to make sure nothing steps out of line. All this limits the benefit of parallelization and discourages programmers from trying too hard. After all, it's a better idea to put more of your effort into areas where performance can be improved more significantly. (John Carmack put it really well once, but I can't remember the quote... and I'm doing too much benchmarking to go look for it now. :-P)
No matter what anyone does, some stuff in the renderer will need to be sequential. Programs, textures, and resources must be loaded up; geometry happens before pixel processing; draw calls intended to be executed while a certain state is active must have that state set first and not changed until completion. Even in such a massively parallel machine, order must be maintained for many things. But order doesn't always matter.
Making more things thread-safe through an extended device interface using multiple contexts and making a lot of synchronization overhead the responsibility of the API and/or graphics driver, Microsoft has enabled game developers to more easily and effortlessly thread not only their rendering code, but their game code as well. These things will also work on DX10 hardware running on a system with DX11, though some missing hardware optimizations will reduce the performance benefit. But the fundamental ability to write code differently will go a long way to getting programmers more used to and better at parallelization. Let's take a look at the tools available to accomplish this in DX11.
First up is free threaded asynchronous resource loading. That's a bit of a mouthful, but this feature gives developers the ability to upload programs, textures, state objects, and all resources in a thread-safe way and, if desired, concurrent with the rendering process. This doesn't mean that all this stuff will get pushed up in parallel with rendering, as the driver will manage what gets sent to the GPU and when based on priority, but it does mean the developer no longer has to think about synchronizing or manually prioritizing resource loading. Multiple threads can start loading whatever resources they need whenever they need them. The fact that this can also be done concurrently with rendering could improve performance for games that stream in data for massive open worlds in addition to enabling multi-threaded opportunities.
In order to enable this and other threading, the D3D device interface is now split into three separate interfaces: the Device, the Immediate Context, and the Deferred Context. Resource creation is done through the Device. The Immediate Context is the interface for setting device state, draw calls, and queries. There can only be one Device and one Immediate Context. The Deferred Context is another interface for state and draw calls, but many can exist in one program and can be used as the per-thread interface (Deferred Contexts themselves are thread unsafe though). Deferred Contexts and the free threaded resource creation through the device are where DX11 gets it multi-threaded benefit.
Multiple threads submit state and draw calls to their Deferred Context which complies a display list that is eventually executed by the Immediate Context. Games will still need a render thread, and this thread will use the Immediate Context to execute state and draw calls and to consume the display lists generated by Deferred Contexts. In this way, the ultimate destination of all state and draw calls is the Immediate Context, but fine grained synchronization is handled by the API and the display driver so that parallel threads can be better used to contribute to the rendering process. Some limitations on Deferred Contexts include the fact that they cannot query the device and they can't download or read back anything from the GPU. Deferred Contexts can, however, consume the display lists generated by other Deferred Contexts.
The end result of all this is that the future will be more parallel friendly. As two and four core CPUs become more and more popular and 8 and 16 (logical) core CPUs are on the horizon, we need all the help we can get when trying to extract performance from parallelism. This is a good move for DirectX and we hope it will help push game engines to more fully utilize more than two or even four cores when the time comes.
109 Comments
View All Comments
ssj4Gogeta - Saturday, January 31, 2009 - link
oh ok. thanks. :)PrinceGaz - Friday, January 30, 2009 - link
I was aware of the idea of the Tesselator in AMD's DX10.1 hardware, but this article was excellent in putting everything together in how it will (or might) function in the DX11 pipeline, and everything else associated with the additions DX11 builds on top of DX10, and the reasons behind what has been done and the implications of it.This is definitely up there with the best of the articles Anand himself posts here; well done Derek. It was the sort of article that first made AT my favourite site to visit first for reviews-- in depth and written by someone who knew what he was talking about. Thankyou.
PrinceGaz - Friday, January 30, 2009 - link
Just because you don't use like it doesn't mean that lots of people are using it day-in day-out on PCs it was pre-installed on. Admittedly I dual-boot Vista and XP, and 99 times out of 100 it is XP (the Vista installation is only used either to help my father when speaking on the phone and who bought a laptop with Vista last year, or to run 3DMark Vantage), but Vista is very popular with most people who use it. My father hasn't any complaints with it (okay so he hasn't used any other O/S, but does that matter).Personally, I intend to jump from XP to Win7 SP1 when it is released. My current Vista installation is 64-bit Ultimate SP1 and it works, but with only 2GB of RAM, I find that XP does everything just as well or better. I would say Vista is very successful though, as almost all computers sold with Windows today have Vista installed rather than XP. There are probably more computers now sold with irrelevant OS's like Linux or OS/X, than there are those that come shipped with XP (netbooks excluded).
PS: actually I like some Linux distros, so if you found the previous paragraph at all offensive, I apologise. But the average user isn't going to want a PC with a Linux distro, unless it comes fully pre-configured with every app they need pre-installed.
ameatypie - Tuesday, February 3, 2009 - link
sorry, but i completly disagree with this. Vista is RUBBISH. If you use it on your average comuter, it is slow. If you use it on a computer worse then that, it is worse then slow - it is useless. If you use it on a fast computer, it wastes half the reasorces that get thrown at it and it doesnt use the other half efficantly enough.The only reason people now days are using vista is because they are forced to with new computers - espically laptops.
I have seen the windows 7 beta, and THAT is going to be pretty good........ But vista is still RUBBISH. :P
just4U - Wednesday, February 4, 2009 - link
No one is FORCED to use Vista and it's certainly NOT rubbish. It is definitely a step way up from XP. I am using the Win7 Beta right now and it's a improvement but why anyone would say its pretty good and Vista is not is beyond me as they are very similiar.I really do think all the Vista bashing has come from people who just don't use it or never really gave it much of a chance. NO instead they go back to (what one user said..) That Bastion of security and stability ... Windows XP. Note the sarcasm. Xp was good for it's time but Vista blows it out of the water.
nilepez - Wednesday, February 4, 2009 - link
By average, do you mean a processor slower than a Athlon 64 3000+ and a 4 year old graphics card?Just curious, because I've got a machine like that that has no problems running Vista (and gets an Vista Experience score between 4 and 5).
Clearly your idea of average is a machine that was, at best, average 4 years ago.
Zoomer - Monday, February 2, 2009 - link
Well, if you throw lots of ram (8GB), fast new graphics card, and use nLite to remove/disable crap like UAC, Defender, firewall, readyboost, and make programs run with administrator privileges by default, it's usable.Nothing like Ubuntu though. However, Ubuntu doesn't run games, or rather, developers don't write mainstream 3D games for it, so I just use VMs or remote connections.
One thing I have not figured out is to change the mouse behavior in windows to more closely resemble linux: mouse wheel should scroll the window the pointer is hovering over, regardless of the active window. Unfortunately, it doesn't seem possible. Therefore, Vista sucks. ;p
gochichi - Thursday, February 19, 2009 - link
Ubuntu is also bloated. I mean, I can't get my hard drive to stop making noises on Ubuntu.. NOR on Vista... which annoys me to no end. I was finally able to stop the noises in Vista but Ubuntu still grinds it away.Ubuntu is the stuff though. I run Office 2003 on it using Wine and that runs great. But the truth is that Office 2007 is actually better and so Vista is better. you start taking games into account and Windows is clearly the best period... consoles and everything... I'll take a Windows box anyday.
Ubuntu is here to stay, it's getting better every six months. Because Windows XP hasn't died and refuses to die... it gives the open source community time to play catch up. Like Wine, it can reach Windows XP status and if it does (when it does, hopefully before Windows XP is totally dead) it'll be a great option.
Ubuntu is always a great option, the problem is that Microsoft OSs are as "free" money-wise because they come preloaded on 99% of PCs. They only people buying separate versions of Microsoft OSs are Apple users (the irony! OH THE IRONY!) and the handful of people builiding their own systems.
The typical case is like my laptop. By default Vista Home Premium... and it does dual boot to Ubuntu sometimes mostly for my Unix class. I think it's a joy to use it either way, but Ubuntu doesn't have any price advantage at all. And it never will because it's cheaper for Dell (in this case) to pay Microsoft the OEM fee for Vista (or Windows 7 in the near future) than it is to offer the technical support on Ubuntu. Vista (even Vista) also gives the system synergy and enhances the value of the hardware which more than justifies the OEM fee. I like dual booting, and I know that I'd get better at Ubuntu if I didn't have Vista but I find dual booting too irresistible.
So Ubuntu is great, it can't be denied. And one of these days Wine (the windows "not" emulator) might catch up enough so that productivity software can run better on Ubuntu than on Windows XP. Windows XP is Microsoft's Achile's Heel.
Because if new Macs didn't run Windows XP in a virtual machine as well as they do, they wouldn't be so widely adopted.
Windows XP symbolizes the overall lack of progress in the industry, and other companies are approaching its usability or swallowing it whole (like through a VW like with Macs and Fusion etc).
Overall, the computer industry has stagnated. Windows 7 might help, and DirectX11 might help, but I think it'll just keep things from slowing down so rapidly. So they'll continue to slow down, just less quickly.
ameatypie - Tuesday, February 3, 2009 - link
but then, how many people apart from myself and a few select others are going to throw 8GB of RAM, an 8800GT or higher series graphics card, and a beefy quad core at it? not many, if any.nilepez - Wednesday, February 4, 2009 - link
1. You don't need that for Vista.2. 8GB of ram can be had for under 100 bucks
3. Vista would run fine on a $20.00 PCI-e card
4. Vista ran fine with a 6600gt or a X800 card (both over 2 years old by the time Vista was generally available).
If you're talking about a card for gaming, well that has little to do with Vista and most tests now show that games run about the same on XP and Vista. This is no different htan XP upon release. It took a year or 2 before games ran about as well on XP as 9x.