Unigine Heaven 2.0 Changes

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
A lot of you wanted to know what the difference is between the new Unigine Heaven 2.0 benchmark we linked yesterday and the old version. Well, aside from an “improved” engine and Linux support, the big addition to version 2.0 is the "extreme" tessellation mode. To give you an idea what “extreme” means, I snapped a few screenshots on my Radeon HD5850 Eyefinity 3x1 setup at 5040 x 1050 (images have been reduced in size to make them easier to download). Notice the difference between no tessellation, normal tessellation and extreme tessellation.
 
Extreme tessellation did nothing to that light pole, therefore it is useless :p.

Isn't "extreme tessellaiton" what Nvidia is touting with the new Fermi cards? If so, I'm all for it since it makes shit more purdy :D.
 
more squiggly lines with extreme tesselation...looks brighter too, should re-name it extreme HDR tesselation
 
Massive fps drop between normal and extreme tessellation with my ati card
 
Looks like there's a massive increase in triangles for extreme mode. Can we have the same screens with the wireframe as well? It'd be nice to also compare how it looks ingame.
 
I'm not seeing much difference outside the wire frame (the actual look of the objects) in the top 2 modes, and it only changes some things, the light pole wasn't touched
 
I wish the benchmark gave the actual numbers of triangles being drawn at a given frame. I think we are seeing things here that are way outside of what we will see in gaming any time soon.
 
Just curious to see how it affected the performance #s for the HD5xxx series.
 
Just curious to see how it affected the performance #s for the HD5xxx series.

performance drop across the board for all ATI users...no coincidence this comes out a few days before Nvidia's Fermi launch...extreme tesselation?...another heavily hyped Nvidia feature
 
There seems to be a serious lack of Depth of Field effects in the overall benchmark too. Several scenes have it removed completely since the first benchmark. I take it the nvidia cards are weaker at that then?

The dragon is just lol with the tessellation. "Moar spinez!!11!11!!"
It's not a dragon any more, it's the nvidia tessellation beast.
 
meh by the time i need to worry about this it will be 3 or 4 generations of cards later

wow now, SC2 later this year, D3 next year.... all i care about really :)

this is just another crap shoot BS selling gimmik like Physx IMO, Nvidia is quickly becoming the new Apple in my books.
 
Light pole and the parts of the building are probably untouched because they're flat surfaces...
 
Of course. Tessellation is barely noticeable in many games and to introduce an extreme mode is cleary Nividia's idea.
They have a GPGPU disguised as a gaming video card so naturally it will excel at Tessellation.
I want to see real side by side in-game performance. Fermi is more of a number cruncher than gaming card.

If the most pesimistic previews have shown the 480 equal to the 5870, and you are saying Fermi isn't a gaming card then I guess the 5870 isn't either?
 
Couldn't agree more; I think the DX11 in this is a few years out from actually making it into a game. However, it does provide a somewhat.... realistic expectation of performance hit when enabling the tessellation. Metro 2033 is surprisingly demanding with Very-High textures and DX11.
What's interesting is that there's not much of a performance hit when enabling tesselation in Metro 2033. With the PC in my sig, playing at 2560x1600 with everything maxed in DX11 except no advanced DoF and AAA anti-aliasing, enabling tesselation is a 10-15% FPS hit at most. It's also tough to notice, mostly seen on the NPC's and the gun models.
 
performance drop across the board for all ATI users...no coincidence this comes out a few days before Nvidia's Fermi launch...extreme tesselation?...another heavily hyped Nvidia feature

nVidia calls every feature they include in a new card "extreme" :rolleyes: The only thing that's going to be extreme with Fermi is the cost...the rest will be "also ran" next to the 5870...6 months late and over priced.
 
Anyone have pictures of the three modes without the wire-frame? I'm curious how big of a difference the extra polygons make to what the end user actually sees.
 
I have a lovely graph of benchmarking results I made yesterday featuring the Extreme mode but I was
using a card that I am still under embargo on. You can see it on my site in a few days if you like. :D
 
The ship in the 90's remake of Lost In Space was made up by 80 million polygons, as a little trivia. How many polygons are pushed by the heaviest of benchmarks today? What tess setting was that Nvidia Fermi bench video running?
 
Further weird things surrounding Unigine Heaven. http://www.overclock.net/8840643-post67.html

If you go into the heaven 2.0 directory, you will find heaven.zip. If you unpack that, you can see all the mesh files. If you have the free SDK, you will find optimized code that looks for specific hardware before defaulting to a secondary engine.

It doesn't name what it is looking for, but it looks for a specific hardware string, and when it is not found, it just falls back to the old render.

We never got to get out hands on 1.1, and Unigine would not send it to me, stating that they did not produce it. I later found out that nVidia bought a license and produced 1.1. As far as I can see, this is not a real world scenario.

And wireframe-mode crashes for 5XXX cards?
 
I don't know how many of you guys have run this benchmark with the latest 2.0 version, but I noticed the wierdest thing. When i'm running Metro 2033 at 1360x768 with DX11, all details on highest setting, DoF on, Tessellation on, and AAA with 16x AF, my GPU's will hit between 84-88degrees.

When I fired this benchmark up, I prepared myself thinking that my GPU fans are going to howl away like during Metro 2033. However, I noticed that my GPU's were only hittin 66 degrees maximum throughout the entire benchmark, and the fans were not howling. Now that just seems wierd, because this benchmark appears to be much more graphically intensive than Metro 2033. I will try to see if I can post my scores later tonight.
 
Further weird things surrounding Unigine Heaven. http://www.overclock.net/8840643-post67.html

And wireframe-mode crashes for 5XXX cards?
Call me cynical, but I'm not very surprised and the release of 2.0 is too coincidental with a certain other release :rolleyes:.

I just downloaded and ran the benchmark with my 5850's at 1000/1250 and got the following:
unigine2runs.jpg

I also benchmarked via FRAPS while running them, here's the FPS graph:
There were some odd dips in the Normal Tesselation run, maybe because I ran it first? Either way, you can see the performance impact from running Tesselation on "extreme." The thing is though, the FPS generally sucks considering the image quality received. Compared to something like Metro 2033, one wonders where the performance goes. I'm going to fire it up again and take some screenshots to carefully examine the tesselation quality.
 
I like seeing the rhythm the tesselation is using to draw more triangles, looks extremely complicated.
 
It is interesting that in the benchmarks and preview videos I've found Nvidia ran this bench with AA off - no wonder their scores were high. Could it be that extreme tesselation clashes a bit with FSAA in the Fermi implementation? Can anyone with a 5870 please run the extreme version without FSAA to get some scores?

Would be interesting which effects have the biggest overall impact of fps, respectively which combination of effects.
 
Back
Top