DX11 Benchmark Results

[DeadlyMACD]

In October Unigine released the first publicly available DX11 Benchmarking program called Heaven. I ran some tests to see what kind of performance my 5770 is capable of and to compare the visual effects to DX10. Here are the results. Tests were run at 1024x768 resolution with Shaders on high, Anisotropic Filtering at 4, and Anti-Aliasing at 4x. Pictures shown are DX10 followed by 11 and showcase areas in which DX11 significantly improved graphical quality. You can download Heaven and several other benchmarking programs at http://unigine.com/

DX10


DX11


DX10


DX11


DX10


DX11


DX10


DX11




Wireframes

DX10


DX11


DX10

DX11


DX10


DX11


DX10


DX11


I ran two other DX11 tests and averaged 33.10 fps with 2x AF/AA and 37.89 fps with no AF or AA.

[Kradath]

Re: DX11 Benchmark Results

Nice screenshots. It really gives me the creeps how got the tesselated surfaces look and how minor the performance drop is compared to what you get.

[jake_]

Re: DX11 Benchmark Results

Nice screenshots. It really gives me the creeps how got the tesselated surfaces look and how minor the performance drop is compared to what you get.

My thoughts, exactly.

[Vreki]

Re: DX11 Benchmark Results

Nice screenshots. It really gives me the creeps how got the tesselated surfaces look and how minor the performance drop is compared to what you get.

Minor? Its a 40% drop. Imagine going from 30 to 20 FPS instead from 49 to 30

ETA: Evidence of bad math skills removed

[jake_]

Re: DX11 Benchmark Results

Minor? Its a 60% drop. Imagine going from 30 to 12 FPS instead from 49 to 30

ur doin' it wrong.

1-(29.8/48.7) = 38.81% drop (61.19% of original). So that would be going from 30-18.36. My value is 152.98% higher than yours. (This is all assuming the FPS change is a linear function).

Of course, that still is a bit of a performance leap, but I would be willing to sacrifice it, since it's actually extremely difficult to tell the difference between decent FPS rates with the human eye.


edit: made the values more accurate for teh mathematical lulz.

[Rambo]

Re: DX11 Benchmark Results

Im still not going to dx11 for some time ,out of discust that dx10 didnt get a chance to be developed.Im not made of money.

even if i won lotto id be thinking this:)

my new 9800GT 1 GB dx10 card should last 2 years?

nice screens tho..

[=98=]

Re: DX11 Benchmark Results

Im still not going to dx11 for some time ,out of discust that dx10 didnt get a chance to be developed.Im not made of money.

No point in developing a technology that never really took off

my new 9800GT 1 GB dx10 card should last 2 years?

Good luck with that!

[Rambo]

Re: DX11 Benchmark Results

well they might release dx12 in 3 months:p

[=98=]

Re: DX11 Benchmark Results

well they might release dx12 in 3 months:p

It would ship with a new OS if anything. Most people could see from the outset that neither Vista nor DX10 were ever going to take off.

DX11 I suspect will be an industry standard for a while now.

[Vreki]

Re: DX11 Benchmark Results

ur doin' it wrong.

1-(29.8/48.7) = 38.81% drop (61.19% of original). So that would be going from 30-18.36. My value is 152.98% higher than yours. (This is all assuming the FPS change is a linear function).

Of course, that still is a bit of a performance leap, but I would be willing to sacrifice it, since it's actually extremely difficult to tell the difference between decent FPS rates with the human eye.


edit: made the values more accurate for teh mathematical lulz.

That was embarrassing I guess the lesson must be that one shouldnt do math in the morning.

But still I think it is a significant drop, considering that it has hardware support. It was also a rather low resolution, it could be even worse at a "normal" resolution.

[Kradath]

Re: DX11 Benchmark Results

Of course the drop is not small if you just look at it. But you have to take at how good the picture looks after that. You could also see it from another way. If you build a game with tesselation so it looks as good as the game without tesselation the performance GAIN shouldn't be minor as well.

[jesus]

Re: DX11 Benchmark Results

Not to mention it's an HD5770. I'd like to see what kind of FPS a 5870 or 5970(?) would get. And these are basically the first DX11 cards. Performance should be even better when the hardware is more "matured"

[Vreki]

Re: DX11 Benchmark Results

It does look good. Maybe DX11 will finally allow the PC to leave the consoles in the dust?

[DeadlyMACD]

Re: DX11 Benchmark Results

The performance was a little disappointing. I'm on a low resolution monitor and frame rates were dropping below 20 in areas with heavy tessellation. The dragon and cottage roof were getting around 17-20 fps and noticeable lag. I'm interested in seeing what kind of performance higher end DX11 cards produce.

[Kradath]

Re: DX11 Benchmark Results

But you have to keep in mind that graphic demos are always looking far better than games of the same generation. They are there to show what is possible and which technologies are new. I think it will be interesting how real games will perform.