So After DirectX 12 looks like there is only slightly improved performance
After DX12
Before DX12

So what about big boost performance increase? Is that happened to you or just marketing boast.

9 years ago*

Comment has been collapsed.

Firestrike is DX11 only.

DX12 won't help non DX12-games/test etc.

9 years ago
Permalink

Comment has been collapsed.

So it only affect DX 12 games? what game for example?

9 years ago
Permalink

Comment has been collapsed.

The API Overhead test in 3DMark Advanced can show what DX12 can do.

9 years ago
Permalink

Comment has been collapsed.

There is no DX12 games (yet).

9 years ago
Permalink

Comment has been collapsed.

Why not share yours.

9 years ago
Permalink

Comment has been collapsed.

because there is no DX12 advantage in the firestrike demo :) and because I have a GTX 750Ti and I'm shy to show you the score :P

9 years ago
Permalink

Comment has been collapsed.

No worries, just to share information not to boast compare.
Too bad can't test DX12 capability though, wonder if my old card can max DX12 too.

9 years ago
Permalink

Comment has been collapsed.

Read about Mantle and other close-to-the-metal APIs. This is what DirectX12 is about. Of course you need DX12 specific games/benchmarks to notice the difference.

9 years ago
Permalink

Comment has been collapsed.

Here are some DX11 vs. DX12 benchmarks for the upcoming RTS game Ashes of the Singularity. The short of it is: DX12 in this game is in most cases a little slower for the GeForce GTX 980, and a lot faster for the Radeon R9 390X. This brings the 390X to the performance level of the 980 in this game.

9 years ago
Permalink

Comment has been collapsed.

That's because the developer nerfed Nvidia cards to further their love affair with AMD. This has been public news since the games Alpha.

9 years ago
Permalink

Comment has been collapsed.

Sounds like NVIDIA fanboy bitching.

If that was true, don't you think they'd have nerfed it for DX11 too?

9 years ago
Permalink

Comment has been collapsed.

Well, this story certainly has no indication that the devs did anything on purpose to nerf NVIDIA cards, and if anything the release of a special driver by NVIDIA indicates that there was a problem with its implementation in the first place.

9 years ago
Permalink

Comment has been collapsed.

I also don´t think ,that this is a intentional nerv. But this is also not a proof of a driver-problem. Possible, but it´s also not uncommon that early alphas on games developed mostly on one kind of hardware, does not run well on other hardware platforms. Of course theoretically DX12 should be faster, but as all previous DX11 vs 10, 10 vs 9 and so on issues, it´s less about performance and more about compatibility when the first games use it.
So my conclusion for this benchmark ist simple....it does not tell us anything about DX12 vs DX11.

9 years ago
Permalink

Comment has been collapsed.

It doesn't tell us about DX12 vs. DX11 in general, but it does tell us that in some cases AMD GPU's gain a lot of performance while NVIDIA ones lose a little. It's certainly possible that optimising for NVIDIA will get better performance, but at this point all we know is the above. At least for AMD that's promising.

It's also clear that they didn't go all out optimising for AMD, since the DX11 scores are much lower than the NVIDIA ones.

If you look at Anandtech's 3DMark API overhead test (which is several months old, so drivers might have changed the results since then), a Radeon 290X can do more DX12 draw calls than a GeForce 980, but when it comes to DX11 the GeForce can do about twice as many calls. So there's definitely something to the relative performances we see in Ashes of the Singularity. It still doesn't tell us why the 980 is getting lower performance in DX12, but otherwise the tests chime well.

9 years ago*
Permalink

Comment has been collapsed.

I actually now agree, maybe I was being a bit over dramatic this morning, from what I've been reading on various media outlets, the developers refused to provide Nvidia with any code until the very last minute due to licensing with AMD. Which In reality says more about the childish way Nvidia and AMD handle their licensing agreements, than the developers themselves.

9 years ago
Permalink

Comment has been collapsed.

3DMark API Overhead feature test:
pix-sauce

the difference will be significant in games as well (depending on the game, settings, general optimization)
tried mantle myself on a weaksauce pc (i5-2400;r9-270x) when it released boosted the star swarm benchmark up to 50% something ...

View attached image.
9 years ago*
Permalink

Comment has been collapsed.

I wanna try that benchmark, downloading :3

Edit:
Results:

View attached image.
9 years ago*
Permalink

Comment has been collapsed.

average fps/total frames increased by 81% - pretty big gains :-) (guessing r9 280x)

9 years ago
Permalink

Comment has been collapsed.

yeah there was a massive increase, no wonder it felt a lot smoother in the second run :) (290x)

9 years ago
Permalink

Comment has been collapsed.

I tested the API benchmark
DX12 10x better, here mine
http://www.3dmark.com/3dm/8241580? - Standard 720p
http://www.3dmark.com/3dm/8241542? - 1080p

DX12 : 12,046,949
DX11 : 1 398 743

about 9x faster

DX12 : R9 290X > GTX 780Ti
DX11 : GTX 780Ti > R9 290X

How about yours?

What is the importance of API in simple word?

9 years ago
Permalink

Comment has been collapsed.

Addition :
3DMark warn not to use API benchmark to compare graphic card.

View attached image.
9 years ago
Permalink

Comment has been collapsed.

What is the importance of API in simple word?

API just means Application Programmer Interface. DirectX is an API (an interface through which programmers can access system/hardware functionality). It's a collection of functions.

The big difference between DX11 and DX12 is, that the overhead of functions has been reduced significantly. Overhead basically means unnecessary data and/or processing. it's something like this:

i have a basket full of apples. i want to get one of the apples.

DX11: i have to take the basket, count the apples, remove one and put back the basket.
DX12: i just take the apple.

9 years ago
Permalink

Comment has been collapsed.

That function not working on non DX12 games, which is not really good situation.
I hope all the game using DX12 later.
Thank you.

9 years ago
Permalink

Comment has been collapsed.

bump

9 years ago
Permalink

Comment has been collapsed.

This man win big
http://www.3dmark.com/3dm/6377330
http://www.guru3d.com/news-story/quick-test-directx-12-api-overhead-benchmark.html

I think that overall quality with DX12, don't use the benchmark to compare graphic card.

9 years ago
Permalink

Comment has been collapsed.

Sign in through Steam to add a comment.