An uncertain moment of truth: Post your Unigine Heaven benchmark score/screenshots

[Apocalypse23]

So I downloaded the Unigine Heaven benchmark to try out my new Asus 5850, with a slight overclock. The demo looks so and so, but I wanted to see what kind of a score I'd get, running Dx11 raped my card heavily, at just 2x AA and 8x AF and a 1440x900 resolution, I noticed that my card struggled horribly. I was looking at fps in the 18-29 range and the only thing that came to my mind was the GTX470/480 series. I know it's just a benchmark, and it doesn't in any way look better than Crysis, but the tessellation and Dx11 raped the 5850 hands down.

You can download the Heaven benchmark at: http://unigine.com/download/

Post your screen shots and your scores of your Unigine Heaven benchmark. You must use at least the following settings:

1440x900 or greater resolution
Windows 7/Dx 11 enabled test
High settings
AA 2x Minimum
AF 8x Minimum

"Looks to Nvidia...but uncertain"

**EDIT**

NEW score using Cat 10.3 a driver:

 

[vj8usa]

Do you fare any better with AA off?

 

[Dark4ng3l]

If you are thinking that Fermi will be significantly better then you should be aware it probably won't be. That video thats on youtube where they compare the GTX480 to a 5870 Makes it something like 40% faster only during the hardest parts of the benchmark and equal in the rest. Now all this is with no AA and low AF, etc. (they obviously picked whatever settings would make them look the best) so i's not representative of what someone with a 400$+ video card would want to use in a game.

 

[Apocalypse23]

Do you fare any better with AA off?

Very similar if not worse than at 2x, the fps was still around 18-30.

 

[Apocalypse23]

If you are thinking that Fermi will be significantly better then you should be aware it probably won't be. That video thats on youtube where they compare the GTX480 to a 5870 Makes it something like 40% faster only during the hardest parts of the benchmark and equal in the rest. Now all this is with no AA and low AF, etc. (they obviously picked whatever settings would make them look the best) so i's not representative of what someone with a 400$+ video card would want to use in a game.

Perhaps, what blows me away is that Heaven doesn't even look that good, it reminds me of what Oblivion looked like and that was in Dx9. Why would Heaven tax the video card so much in Dx11, just because of tessellation? And also, do we really need tessellation if we can just create the same effects using standard 3d technology? Oblivion looked not very different than this demo....

Or am I totally wrong here?

 

[Dark4ng3l]

Perhaps, what blows me away is that Heaven doesn't even look that good, it reminds me of what Oblivion looked like and that was in Dx9. Why would Heaven tax the video card so much in Dx11, just because of tessellation? And also, do we really need tessellation if we can just create the same effects using standard 3d technology? Oblivion looked not very different than this demo....

Or am I totally wrong here?

Apparently heaven is way way overboard with tessellation. I read somewhere that close up to the dragon it tessellates so much that some polygons are smaller than the individual pixels that end up on your screen. Pretty much they went way overboard and they used that feature beyond what you would reasonably see implemented in a real game (they went WAY past the point of diminishing returns)

 

[Apocalypse23]

Apparently heaven is way way overboard with tessellation. I read somewhere that close up to the dragon it tessellates so much that some polygons are smaller than the individual pixels that end up on your screen. Pretty much they went way overboard and they used that feature beyond what you would reasonably see implemented in a real game (they went WAY past the point of diminishing returns)

Ah...*curses nvidia*

 

[AndroidVageta]

Perhaps, what blows me away is that Heaven doesn't even look that good, it reminds me of what Oblivion looked like and that was in Dx9. Why would Heaven tax the video card so much in Dx11, just because of tessellation? And also, do we really need tessellation if we can just create the same effects using standard 3d technology? Oblivion looked not very different than this demo....

Or am I totally wrong here?

I would say that tessellation is an improvement over something like parallax mapping, but on something like cobble stone roads, I dont think your average to high end player would notice the difference. Even in games like STALKER Call of Pripyat, sure the tessellation on the characters is nice, but unless its a game like Mass Effect 2 where you can really see the characters up close (and have to when talking to them) a game like STALKER is kinda using it to the extent of wasting resources...I didnt notice that tessellation was on because I always shot them from afar.

And plus, a lot of programs (like the Heaven benchmark) use tessellation to the extreme...I promise you that the amount of triangles produced on screen is 1000% more than would actually be used in any REAL game.

 

[happy medium]

31.4 fps
Score 791

1280x1024

Same settings as yours except no aa

 

[sandorski]

No AA, 4 AF 1024x768 Windowed(missed the Full Screen box, didn't have the patience to do it again) DX9, WinXP Pro 64bit

 

[SolMiester]

No AA, 4 AF 1024x768 Windowed(missed the Full Screen box, didn't have the patience to do it again) DX9, WinXP Pro 64bit

But the object of the post was to measure tessellation, i thought that was only available on dx11?

 

[sandorski]

But the object of the post was to measure tessellation, i thought that was only available on dx11?

Was that the Object? If so, my bad, I do believe you are correct that DX11 is the only way to use that feature.

 

[bunnyfubbles]

My first DX10 card (the 8800GT) was/is too slow to run the DX10 games at settings I want to run, I never bought that card with the specific intent of running DX10 games.

The same is true of my 5850 and DX11.

 

[Headfoot]

Ah...*curses nvidia*

what????? The benchmark isn't made by nVidia, its made by a 3rd party.

 

[SlowSpyder]

1920x1200
2xAA
16xAF
Vista 64

32FPS / 807 score.

OP, I also ran my 5870 at your settings (144x900, 2xAA, 8xAF) and got 42.7FPS / 1076.

That's with my 5870 at 910/1250 (and climbing )

 

[NoQuarter]

If you are thinking that Fermi will be significantly better then you should be aware it probably won't be. That video thats on youtube where they compare the GTX480 to a 5870 Makes it something like 40% faster only during the hardest parts of the benchmark and equal in the rest. Now all this is with no AA and low AF, etc. (they obviously picked whatever settings would make them look the best) so i's not representative of what someone with a 400$+ video card would want to use in a game.

Also the Fermi results are supposedly on 1.1 (2.0) of the Heaven benchmark with aggressive culling that improves performance up to 30% over the 1.0 version available to public so can't compare the numbers straight across to what we get.

 

[zagood]

I don't have a DX11 card, but general question...don't you want a benchmarking program to kick your card's ass? I mean isn't that the point? It's not like you're getting a low framerate that keeps you from playing an awesome game.

Also since DX11 is so new and Heaven is pretty much the only recognized tesselation benchmark, we can't really say where and why the slowdowns are going to translate into actual games.

 

[Sentry2]

1440x900 2xAA 8xAF High settings Win7 Pro/ DX11

FPS: 46.1

Scores: 1162

i5-750 @ 4.0GHz, ASUS 5850 @ 1040 gpu / 5060 mem

 

[StinkyPinky]

I got 27.1 fps at 1680x1050 with 4xAA and 4xAF High Settings.

Tessellation looks great by the way. Never seen it until this demo.

 

[Rubycon]

I don't trust the frame counter. Despite it being well over 60fps the whole time the rendering looked choppy to me. I tried everything maxed out at 2560x1600 and it barely hits 2fps. FAIL!

 

[SlowSpyder]

I don't trust the frame counter. Despite it being well over 60fps the whole time the rendering looked choppy to me. I tried everything maxed out at 2560x1600 and it barely hits 2fps. FAIL!

Yea, one thing I noticed was that no matter what the frame rate, I could see some 'hitching' here and there... like there's be an instant where the rendering would get choppy despite the FPS showing a rate that should look smooth.

 

[GaiaHunter]

If you are thinking that Fermi will be significantly better then you should be aware it probably won't be. That video thats on youtube where they compare the GTX480 to a 5870 Makes it something like 40% faster only during the hardest parts of the benchmark and equal in the rest. Now all this is with no AA and low AF, etc. (they obviously picked whatever settings would make them look the best) so i's not representative of what someone with a 400$+ video card would want to use in a game.

Careful - the GTX480 is running the Unigine Heaven version 1.1 not the 1.0 that is available to public.

 

[Keysplayr]

With a 5870 in the rig in my sig, at 1920x1200 all settings maxxed, I
Get a score of 682 at 22.2 fps. This card does take a serious hit turning tesselation on. Example would be the dragon in the courtyard scene. Tesselation off, I get around 44 fps. Turn it on, framerate drops to 15-17 fps. I'll post screenshots later. On my blackberry right now.

 

[Keysplayr]

If you are thinking that Fermi will be significantly better then you should be aware it probably won't be. That video thats on youtube where they compare the GTX480 to a 5870 Makes it something like 40% faster only during the hardest parts of the benchmark and equal in the rest. Now all this is with no AA and low AF, etc. (they obviously picked whatever settings would make them look the best) so i's not representative of what someone with a 400$+ video card would want to use in a game.

Bold ^
Ermm... Isn't that what you might want? A card that holds up when the going gets rough?

 

[SlowSpyder]

Bold ^
Ermm... Isn't that what you might want? A card that holds up when the going gets rough?

I've read a few different places that Nvidia didn't use AA for the test, assuming that is indeed the case, you have to wonder why that is. Most people who buy this level of video card will want AA. Maybe the going gets a little rough for Fermi with AA in this test? Maybe it's not true, they did use AA? Maybe something else? I'm sure Nvidia will use the settings that showcase Fermi in the best light.