Didn't notice that the 1080 lost performance. Starting to think some mockery is being done here. Yes the Nvidia cards perform poorly in DX12 compared to AMD, but not this poorly. I've seen this the other way around. Most likely driver updates would bring DX12 in line with DX11.
Last edited by Vash The Stampede; 2016-09-08 at 07:55 PM.
Developers have said the performance decrease on 1070/1080 cards is due to a bug that is meant to be fixed.
I've had something like that (well exactly) found out that it was actually caused by shadowplay + monitor capture. Might not be the case for you but it was like this when I was in window mode. You can see the FPS counter at the top left really doesn't match the frame rate being displayed.
guru3D results are using the built-in benchmark
these are the actual results using regular gameplay post-DX12 patch from computerbase.de (pretty much the most competent and thorough benchmarking site around atm):
and yeah the game is Gaming Evolved so it should favor AMD, but it seems the current implementation of DX12 is equally bad for both vendors
good thing though with Nvidia you pretty much dont need DX12 at all right now
Last edited by Life-Binder; 2016-09-13 at 08:31 AM.
Not sure what's going on there with those benchmarks.
Deus Ex is broken in DX12 for both Nvidia and AMD, funnily enough it's actually worse on AMD cards:
http://techreport.com/review/30639/e...kind-divided/3
Completely illogical conclusion. First of all, multiplayer FPS games are always more performance optimized than any other kind out there, there is no point in it if you cant play it at a very high FPS (at least 120). Second of all, Doom uses Vulkan API and Deus Ex uses DirectX 12, and we have no idea how easy that is to optimize game performance in those two compared to each other.
R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B
What does this have to do with Deus Ex which is a single player game? Yes Doom has multilayer functionality, but that's not really why it does so well with Vulkan. After John Carmack left, they hired Tiago Sousa who worked on CryEngine.
My point is that we've seen games that use both DX12 and Vulkan poorly. The Talos Principle for example is a Vulkan game that implemented the feature within a weeks worth of time, and has nearly 0 gains in performance for both AMD/Nvidia. Bad implementations of either API can cause a performance loss. Games built from the ground up for the API tend to do better, and Deus Ex doesn't seem to be like that. More than likely AMD forced DX12 down the developers throats and weren't prepared to adjust the game to benefit from it.Second of all, Doom uses Vulkan API and Deus Ex uses DirectX 12, and we have no idea how easy that is to optimize game performance in those two compared to each other.
Also DX12 for Deus Ex is in preview not final. Even the TechReport article mentions this.
'It's also clear that it's too early to call a winner between the green and red teams for DirectX 12 performance in this beta build of Deus Ex, even if AMD seems to feel confident in doing so."
A quick look for the latest benchmarks of this game with DX12 shows that other people are getting different results. This guy here who released this video 18 hours ago from me making this post, shows the typical scenario of AMD gaining in performance. Why is his results different than others? I don't know.
Single player game developers dont have an incentive to optimize the game that much and usually focus more on the picture quality than FPS. Also, the difference could be engine specific: engines made for a single player game often have to be seriously reworked to optimize performance. We also dont know how "native" is DirectX/Vulkan support is in those new games, simply because these API are new and some of those games have been in development long before those API have released.
That's exactly what I'm talking about, I agree. However we have no idea how that API support was implemented and why does a specific GPU have an advantage. It's also completely unfair to say that Deus Ex engine team is less competent (which is what I feel you're implying) simply because Id team has those achievements behind their back.
R5 5600X | Thermalright Silver Arrow IB-E Extreme | MSI MAG B550 Tomahawk | 16GB Crucial Ballistix DDR4-3600/CL16 | MSI GTX 1070 Gaming X | Corsair RM650x | Cooler Master HAF X | Logitech G400s | DREVO Excalibur 84 | Kingston HyperX Cloud II | BenQ XL2411T + LG 24MK430H-B
The only "proper" DX12 implementations I can think of is Ashes, Time Spy and maybe Warhammer.
I wouldnt call timespy good either... With all the controversy surrounding that one.
I think it is time for a proper game/benchmark to be developed with only DX12 in mind, no DX11 codepath. With all the features of DX12, not taking in account whether AMD or Nvidia can run it. But leave options in there to turn stuff off... I guess that would be easier in a benchmark format. Say, focussing benchmark 1 on that feature A, benchmark 2 on feature B etc. And than a final benchmark all combined...
I guess I can dream...
The only reason Time Spy is a issue is that Pascal gets some performance boost with Async, despite Ashes doing the exact same thing. Time Spy is pure DX12 and the only reason it is not on the highest DX12 feature level is due to the fact that most Radeon cards won't be able to run it, making it pretty useless as a benchmark.
Every released and upcoming DX12 game uses the basic DX12 level anyway.
Interesting current events, it seems that Nvidia is closer to their MSRP price than AMD is. Which considering Nvidia's Founders Edition Fiasco, you would think Nvidia would be last to reach their MSRP. AMD has a history of pricing problems with their cards, and this time it doesn't seem to be any different. Probably due to people using them to BitCoin farm, or whatever they use now. But there needs to be some pricing control, cause if gamers aren't buying these cards it isn't going to do them any good. If Steam's Hardware survey shows Nvidia is still gaining market, these bitcoin miners are going to hurt AMD in the long run.
http://store.steampowered.com/hwsurvey/directx/
for Vulkan systems
1080 - 0.94%
1070 - 1.36%
1060 - 0.48%
480 - 0.26%
470 - 0.01%
460 - 0.02%
Last edited by Life-Binder; 2016-09-17 at 01:02 PM.