Unigine Heaven 2.0 RC1

[96redformula]

More here!

That might deserve a thread of it's own IMO....

[BrowncoatGR]

More here!

a) It's just a benchmark/tech demo for the engine not something that would make it into any game.
b)There's no proof of anything here. It could be Nvidia(or AMD for that matter) put some dev work into the engine to give it an optimized renderer. Either way we don't know atm.

Any way to tell what mode is being used on specific cards or are we relying on what the engine programmer is telling us?

You're probably relying on the cards driver actually

[Eastcoasthandle]

Unfortunately, since we now know that Nvidia licensed 1.1 for Fermi specifically (which is why no one was actually able to get a copy from the internet) we have to treat 2.0 as suspect and wait and see how it performs in the reviews.

[blindbox]

From your link, it seems like unigine is a good test to see whether your memory is clocked too high or not for Radeon HD 5xxx series, by checking the min fps.

[zalbard]

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.

wtf?

[Blacklash]

Done on my AMD rig.

1920x1200 DX11 default filtering with normal tess = 40.8. Link to a full detailed screen cap below.

http://img144.imageshack.us/img144/6420/heavenv2.jpg

Card was a single HD 5870 @ 1030|1300 using the 10.3a driver. If I enable replication in the console my avg FPS increases by 6 to 46.8.

Sans tess, aka current nVidia style, I do 61FPS avg at the noted settings and res. That's without replication on. Didn't bother switching it on for the no tess bench.

[NaMcO]

From your link, it seems like unigine is a good test to see whether your memory is clocked too high or not for Radeon HD 5xxx series, by checking the min fps.

You mean if the clock is too high, the GPU has to wait for it in order to fetch/send data, right?

That would be useful for a fine tune indeed

[Aerwidh]

You mean if the clock is too high, the GPU has to wait for it in order to fetch/send data, right?

That would be useful for a fine tune indeed

I think they mean that if your memory is clocked too high, you get so many errors that the amount of error correction actually lowers performance compared to if you ran the memory slower without much errors.

[Mungri]

I did an error checking test for my vga ram clocks at 1300, 1375 and 1445 Mhz:



It seems to be fine

[BrowncoatGR]

wtf?

They bought a license and produced their own version of the benchmark it seems. I don't understand how that puts Unigine in a bad light.

[bill_d]

They bought a license and produced their own version of the benchmark it seems. I don't understand how that puts Unigine in a bad light.

well they sold out, put a nvidia logo on it and be done with it

[zalbard]

They bought a license and produced their own version of the benchmark it seems. I don't understand how that puts Unigine in a bad light.

It loses its credibility. Who knows what they did to it to make Fermi look better or ATI look worse.

[flippin_waffles]

So will nv be distributing it's version to reviewers and demand in their review guidelines that it be used, or are reviewers going to use the official version with the key given out by AMD to reviewers? Or am I reading this wrong?

[Eastcoasthandle]

Why not? What other reason to make a benchmark specifically tailored for your hardware? They already used 1.1 once in their video. It will be up to the reviewer to say if they are using 1.1 or not. However, if they didn't tell you how would you know? The real question we need to ask is if 2.0 is a derivative of 1.1? Or, is 2.0 simply 1.1 with some minor tweaking?

[RaZz!]

Tesselation works on 4850 with opengl mode

So it's using 4-series fixed-mode tesselator

how? just ran heaven 2.0 with my 4850 (on win7 64bit). i can't enable tessellation in opengl.

[BrowncoatGR]

how? just ran heaven 2.0 with my 4850 (on win7 64bit). i can't enable tessellation in opengl.

Need to have the RC or run it under linux(linux tesselation performance is horrible though). Or try editing the config file. Haven't tried that on windows, but it should work.

[Overfiend]

just thought I'll post my scores

@1920x1200 - AA4,AF16,DX11,all high, Tesselation Extreme

score: 1048
avg: 41.6

min: 18.0
max: 93.4


rig: w3520@4GHz - 2x5870 Crossfire - 12GB

[koc]

Just installed the Heaven Benchmark 2.0 and run some tests

i7 920 @4GHz - 3x280 Tri Sli @stock

@1920x1200 - AA8,AF16,DX11,all high, Tesselation Extreme

FPS: 58.6
Scores: 1476
Min FPS: 22.7
Max FPS: 118.6

i7 920 @4GHz - 3x280 Tri Sli @675/1453/1175

FPS: 64.5
Scores: 1625
Min FPS: 24.5
Max FPS: 130.5

[Florinmocanu]

so... 3 x GTX 280 = DX11 and tesselation? I think you should re-check what you are typing, there's no DX 11 support and tesselation support in GTX 280 video cards.

[koc]

so... 3 x GTX 280 = DX11 and tesselation? I think you should re-check what you are typing, there's no DX 11 support and tesselation support in GTX 280 video cards.

Sorry I copy the sentence from Overfiend reply and i forget to edit it

DX10
tesselation disable

[NapalmV5]

stock cooler

[NapalmV5]

stock cooler

[NapalmV5]

stock cooler

[Pontos]

New version of three Unigine benchmarks @ http://unigine.com/
Main addition being OpenGL 4.0 (Thus Tessellation) on both Windows and Linux, and Nvidia 3D Stereo on the Heaven v2.1 bench.


Fast bench on Tropics D3D10 and OpenGL with mah crappy PC which somehow has a CPU running in THz in one bench, and in negative THz on another
Real speed is on my sig.



OpenGL path takes a noticeable hit... I may try later on my Linux installation.