Is DirectX 12 Worth the Trouble?

[Rroff]

Developers still complain that dx 11 is too restrictive etc. yet they still use it (granted not exactly much other choice for them). From what I can see and what the original article says, it just comes across like the developers aren't familiar with dx 12 and that they need more time/practise to be comfortable working with it in order to get the best from it and given how hard developers are pushed to meet deadlines already and as a result not even being able to finish the game for release resulting in day 1 patches and many other patches throughout the first month or so, this means that they can't get to proper grips with it.

Maybe, for all we know, we could see a dx 12.1 or even dx 13 before there are any real dx 12 titles but I be willing to bet that whatever the next version is, it will just be an improved version of dx 12 as I can't see microsoft releasing something completely different to dx 12 so soon.

It isn't just about familiarity - generalising a bit but look up a function in DX11 and there will usually be like a 3 line description and you can more or less call that function with a couple of parameters/pointer to a buffer and forget about it while with DX12 that function probably doesn't exist and instead you will find a lengthy description of the concept of what that function does and how to implement it along with flowcharts and require significant support use of data mapped into memory before and after amongst other things.

 

[smogsy]

a lot of people who forget that with DX12 being so low level that everything needs to be managed correctly to see "ok" performance. with DX11 DX handles a large amount of the management making it easier to develop and get right.

 

[Rroff]

Yeah that is true - very easy to screw something up in DX12 and get poor performance while despatching it indiscriminately in DX11 would still get ok performance.

 

[Nexus18]

It isn't just about familiarity - generalising a bit but look up a function in DX11 and there will usually be like a 3 line description and you can more or less call that function with a couple of parameters/pointer to a buffer and forget about it while with DX12 that function probably doesn't exist and instead you will find a lengthy description of the concept of what that function does and how to implement it along with flowcharts and require significant support use of data mapped into memory before and after amongst other things.

No doubt some parts will need to be tweaked, I expect to see improvements for the likes of the above in dx 12.1 or something down the line.




As for dx 12 being worth it or not... I just got deus ex: MKD so not properly played with settings etc. (mostly on high with no MSAA) but once again dx 12 is providing a nice wee boost for my system with a consistently lower frame latency:

 

[Kaapstad]

Well you could argue that since dx 12 significantly improves performance/min fps for certain people, it allows them to increase graphic settings therefore you are in some ways getting better IQ.

As DX11 and DX12 look exactly the same in all the games I have seen there is no image improvement.

 

[Kaapstad]

A whole load of stuff that does not work? ONE thing that developer mentioned, ONE thing, that was stacked memory, and only complete idiots were saying it was going to be a big deal. I specifically and repeatedly said it won't happen. It's non viable to access memory with which you need 200GB/s speed access across a 16GB/s bus which is already in use.

So the bottom line is, after reading it, you made up a bunch of things that don't work, when they didn't say that at all.

We weren't supposed to get fantastic image improvements either, once again you just made that up. Though it's worth noting that several game reviews have showed superior quality shadows and things under DX12 than 11.

DX9, DX10, DX11, DX12, Vulkan and Mantle do not offer superior image quality than the last, they offer superior performance, and more tools to ENABLE devs to add higher quality graphics if they choose.

If your game runs at 60fps in DX11 and 70fps in DX12, then you might not add a graphical effect that costs 10fps to the DX11 game, but you might choose to add it under DX12, in fact it might only cost 5fps to add it under DX12, thus if your target was 60fps in that game on specific hardware, you can meet it in DX12 but not 11.

There is not much you can do in a new API that you can't do in an old API without work arounds or higher performance cost. It's interesting that throughout your attacks on DX12(and anything AMD) it's always stuff like this, just random stuff you say with no proof.

You posted an article in which a dev specifically stated multiple ways in which DX12 was a direct improvement and had features that were the future of programming games.... then followed it by saying it was a white whale, ie, it's worthless. Anyone that can post an article with a dev stating DX12 is a direct improvement, but claims to read it's worthless, did not read the article or has an agenda to misrepresent the article.

Apart from the lack of image improvement (which I did not make up) how about the lack of mGPU support to mix different cards.

Worse still my Maxwell Titans in 4 way SLI using DX11 is a faster gaming setup than anything you can get with Pascal, answer because DX12 has wreaked mGPU support. NVidia can not be bothered with any more than 2 way Pascal support and AMD don't have anything on the market that can justify 4 way support even if they wanted to bother.

 

[Nexus18]

As DX11 and DX12 look exactly the same in all the games I have seen there is no image improvement.

Never said that there was any difference solely based on DX 11 vs 12, I said with DX 12 giving better performance, it allows people to turn settings up thus in some ways you are getting better IQ than what you could achieve with DX 11.

 

[Bazzards]

Tried dx12 and now recall why I don't use it, no sli.

 

[Kaapstad]

Never said that there was any difference solely based on DX 11 vs 12, I said with DX 12 giving better performance, it allows people to turn settings up thus in some ways you are getting better IQ than what you could achieve with DX 11.

So why does DX11 run faster on NVidia cards.

Please don't say that AMD cards are better optimised for DX12 as this means nothing when NVidia have the fastest DX12 cards.

 

[Nexus18]

So why does DX11 run faster on NVidia cards.

Please don't say that AMD cards are better optimised for DX12 as this means nothing when NVidia have the fastest DX12 cards.

Lol...

Not sure if serious... If so then

*Facepalm*

If you can't see that AMD cards gain a lot more from DX 12 than what nvidia cards do then there is no point me trying to explain that to you.

As for why DX 11 is better than DX 12 for nvidia users, maybe you should ask nvidia for an update on an ETA for when their promised DX 12 optimisation driver will drop...

Let's just say that I'm very glad I chose a 290 over a 780/970.

The fact that you expect developers + AMD + nvidia to provide great 4 way support says it all, explain why they should waste resources for an extremely small minority? Heck even though nvidia said that they wouldn't be supporting 4 way sli for pascal and gaming, you still bought 4

 

[Overlag]

DX12 is worth it. Adoption takes time. This article was probably written by Nvidia.

 

[Mauller]

Dx12 is worth it when games start taking full advantage of its abilities. Ofcourse at the moment the differences may be minimal for someone running a very high end rig, but that is only because every game released with DX12 so far has been designed to run within the limitations of DX11.

Once games start becoming more dynamic, expansive and with more unique objects on screen, then DX12 will start coming in to its own. There are plenty of titles already that would benefit from DX12, most of them open-world games. The reason most games don't see advantages is becasue they are corridor based shooters at most with very few unique objects on screen.

 

[Kaapstad]

Lol...

Not sure if serious... If so then

*Facepalm*

If you can't see that AMD cards gain a lot more from DX 12 than what nvidia cards do then there is no point me trying to explain that to you.

As for why DX 11 is better than DX 12 for nvidia users, maybe you should ask nvidia for an update on an ETA for when their promised DX 12 optimisation driver will drop...

Let's just say that I'm very glad I chose a 290 over a 780/970.

The fact that you expect developers + AMD + nvidia to provide great 4 way support says it all, explain why they should waste resources for an extremely small minority? Heck even though nvidia said that they wouldn't be supporting 4 way sli for pascal and gaming, you still bought 4

Show me an AMD single GPU card that can run DX12 faster than the top NVidia cards, no you can not.

Q.E.D.

 

[Rroff]

Once games start becoming more dynamic, expansive and with more unique objects on screen, then DX12 will start coming in to its own. There are plenty of titles already that would benefit from DX12, most of them open-world games. The reason most games don't see advantages is becasue they are corridor based shooters at most with very few unique objects on screen.

Admittedly it would probably run a lot faster ported to Vulkan or something - but back in the day City of Heroes could throw an insane amount on screen and still run acceptably - largely due to good programming and some innovative approaches - IMO ability to think outside the box and approach problems from methods not on the taught familiar path is holding back the level of detail, etc. on screen more than the API is.

 

[Nexus18]

Show me an AMD single GPU card that can run DX12 faster than the top NVidia cards, no you can not.

Q.E.D.

Lol, really, you are comparing 2+ year old GPUs to nvidia's current gen? Not to mention the price difference.... I don't need to say anymore How about you show me some results where nvidia actually gain something from dx 12:

A 290x matching a 980ti:

http://www.guru3d.com/articles_pages/hitman_2016_pc_graphics_performance_benchmark_review,7.html

A fury x beating a 980ti and matching a 1070:

https://www.overclock3d.net/reviews..._the_division_directx_12_performance_review/2

Yup, nvidia cards are so good at dx 12 that they even lose performance Nvidia will drop the miracle driver any day now *stillwaiting.jpg*

And you can google the rest of the dx 12 games to see similar results for amd cards and dx 12.

It will certainly be interesting to see where the likes of the 290, fury x etc. fall come dx 12 only titles, lets hope for yours and the other naysayers sake, that we don't see a repeat of the current situation where a 290 matches and even beats a 780ti/980 in certain things...

Q.E.D.

DX12 is worth it. Adoption takes time. This article was probably written by Nvidia.

It certainly looks that way especially when the OP finishes with:

now it is starting to look like a white elephant

But when you read the article, they are actually praising dx 12 and saying that if you put the time into it, it is worth it. Drunkenmaster has pretty much torn apart kaap's argument for dx 12 not being worth it

https://forums.overclockers.co.uk/threads/is-directx-12-worth-the-trouble.18770873/#post-30549855

Someone just isn't pleased because multi gpu support is non-existent lately thus they can't be topping the game benchmark threads with their 4 way setups lol... Now if dx 12 had superior multi-gpu support for 4 way, you can be guaranteed that someone would be singing from the roof tops about it

 

[Kaapstad]

Lol, really, you are comparing 2+ year old GPUs to nvidia's current gen? Not to mention the price difference.... I don't need to say anymore How about you show me some results where nvidia actually gain something from dx 12:

A 290x matching a 980ti:

http://www.guru3d.com/articles_pages/hitman_2016_pc_graphics_performance_benchmark_review,7.html

A fury x beating a 980ti and matching a 1070:

https://www.overclock3d.net/reviews..._the_division_directx_12_performance_review/2

Yup, nvidia cards are so good at dx 12 that they even lose performance Nvidia will drop the miracle driver any day now *stillwaiting.jpg*

And you can google the rest of the dx 12 games to see similar results for amd cards and dx 12.

It will certainly be interesting to see where the likes of the 290, fury x etc. fall come dx 12 only titles, lets hope for yours and the other naysayers sake, that we don't see a repeat of the current situation where a 290 matches and even beats a 780ti/980 in certain things...

Q.E.D.



It certainly looks that way especially when the OP finishes with:



But when you read the article, they are actually praising dx 12 and saying that if you put the time into it, it is worth it. Drunkenmaster has pretty much torn apart kaap's argument for dx 12 not being worth it

https://forums.overclockers.co.uk/threads/is-directx-12-worth-the-trouble.18770873/#post-30549855

Someone just isn't pleased because multi gpu support is non-existent lately thus they can't be topping the game benchmark threads with their 4 way setups lol... Now if dx 12 had superior multi-gpu support for 4 way, you can be guaranteed that someone would be singing from the roof tops about it

I have got plenty of AMD cards but if I wanted to run DX12 it would have to be on NVidia hardware.

It is not my fault that the best AMD can do is 2 year old cards, that is down to AMD.

 

[tommybhoy]

Nvidia DX12, cutting their nose off to spite AMD's face even at the detriment of their own users...

 

[drunkenmaster]

Apart from the lack of image improvement (which I did not make up) how about the lack of mGPU support to mix different cards.

Worse still my Maxwell Titans in 4 way SLI using DX11 is a faster gaming setup than anything you can get with Pascal, answer because DX12 has wreaked mGPU support. NVidia can not be bothered with any more than 2 way Pascal support and AMD don't have anything on the market that can justify 4 way support even if they wanted to bother.

A yes you did make up image improvement, NO API IMPROVES THE IMAGE DIRECTLY. None, ever, it helps brings up performance so you can if you want to, add higher image quality settings, simple as that.

Second, multi gpu with different brands of cards DOES work and I believe is used in AoS. But it's also pointless, who is buying two different brands of gpu to have in the same system no one. But where stacked memory literally doesn't use because it's a pipe dream, mgpu with different cards has been proven to work, it's just equally as pointless because no one will ever use it.

As for the rest, I'm not sure what the hell you're attempting to bring into the discussion here. 4 Maxwell's are faster, because Nvidia stopped supporting 4 way Pascal.... which is because DX12 sucks????? Really? Nvidia realised like 8 people in the world bothered with more than 2 Nvidia cards in a system, the cost of supporting it and trying to get devs to get scaling beyond 2 cards simply wasn't offering them any payback so they stopped. It has precisely nothing at all to do with DX12.

I know, DX12 is so bad that global warming is getting worse, the rich are getting richer, Trump got elected and child slavery still exists. Is there anything else that has nothing to do with DX12 that you want to blame on DX12?

 

[Rroff]

Second, multi gpu with different brands of cards DOES work and I believe is used in AoS. But it's also pointless, who is buying two different brands of gpu to have in the same system no one. But where stacked memory literally doesn't use because it's a pipe dream, mgpu with different cards has been proven to work, it's just equally as pointless because no one will ever use it.

Explicit multi adaptor does work and doesn't need to be different cards and/or if programmed for properly can work seamlessly with different configs - potentially its very useful if a game can farm out its work load discretely - even just splitting up rendering of the game scene and UI to different GPUs could provide a fairly significant performance boost in some games. Combing VRAM in any useful way for realtime sensitive applications however as you said is a long way off and requires a completely new generation of hardware.

 

[Sampsy]

I thought the whole idea of dx was to make game development easier, I think dx12 is a step backward In that regard.

I think this is the main reason why there really aren't many good implementations. Lower-level access to the hardware is very helpful if you want to improve efficiency but it requires a greater level of expertise and development time. Because of that for everyone apart from the very biggest developers it just isn't worth the effort over an easier to use higher-level api.

Personally I have yet to play any games in DX12. I have tried it out a few times but on every occasion DX11 / the alternative outperformed it.