Unigine 'Heaven' - The First DirectX 11 Benchmark

[Apoptosis]

Unigine 'Heaven' - The First DirectX 11 Benchmark

Last week on October 22, Unigine Corp. announced the release of the World's first true DirectX 11 benchmark "Heaven" that is based on its proprietary Unigine engine. This is fairly big news as we can finally get to see how DirectX 11 will look and run on the latest DirectX 11 graphics cards from AMD. Read on to see how the Radeon HD 5870, 5850, 5770 and 5750 do on this tough DX11 benchmark!

The Unigine 'Heaven' DirectX 11 benchmark looks really impressive, so before we go straight to the benchmark numbers we wanted to show you some screen shots that we took from the benchmarThe scene above is rendered in DirectX 9 and you can see how the roof and stone work in the building corners is relatively flat. This scene reminds me a little bit of the cs_italy map from counter-strike, which is sad considering that map was released over seven years ago. With DirectX 11 enabled the same scene comes to life thanks partly to Tessellation, which is one of the more important DirectX 11 features. Without Tessellation there is no displacement so things are flatter in DirectX 9 or with Tessellation turned off.

Article Title: Unigine 'Heaven' - The First DirectX 11 Benchmark
Article URL: http://www.legitreviews.com/article/1117/1/

[DL13]

Do you know the price for it?

[xthefenixx]

Could you do a test using crossfire?

[Apoptosis]

Do you know the price for it?

For what? The benchmark itself is free - http://www.unigine.com/download/

[Apoptosis]

Could you do a test using crossfire?

It takes ~ 4.5 minutes for the test to run and I benchmarked 4 cards at 4 resolutions/settings each 3 times to get an average of the scores.

That would be 48 runs at 4.5 minutes each or roughly 216 minutes Then I did the API runs, which is 12 more runs or 54 more minutes... I have 4.5 hours tied up in benchmark just these four cards... If I did CrossFire testing that is 4.5 more hours... I really just don't have the time right now.

This was just a for fun article and was posted on the weekend. If this article brings in big traffic and makes it worthwhile to run more tests I'll do a CrossFire update. Not trying to complain, but I always get the 'do more tests' and 'why didn't you do this' and people don't understand the amount of time involved to generate a couple charts. If it was automated it would be one thing, but this requires sitting in front of the screen for hours on end.

[DL13]

Do you know the price for it?

For what? The benchmark itself is free - http://www.unigine.com/download/

Cool, All the other benchmarks you pay for, and this is the first time I seen this one.

but this requires sitting in front of the screen for hours on end.

And some times to the point where your eyes pop out, and that's not fun.
Thanks for the hard work too.

[dgood]

might I ask what version of opengl it is native for? 2.0? 2.1? 3.0? 3.2? However I wouldn't mind seeing some pictures if it is 3.2. Do graphics cards need to be designed for support of opengl 2.0, 3.0 etc like DX versions? Does the 5xxx series AMD/ATI cards support the newest opengl?

[DMB2000uk]

Cool, I'd seen the demo video, but had no idea they'd realeased it as a benchmark too. Thanks for the writeup.

Dan

[imsochobo]

Hi.

There is a known fact that anythng dx can do opengl can do, in most cases that is.
'
Does opengl support tesselation and all that? did opengl just have better performance at same img quality ?

[Apoptosis]

might I ask what version of opengl it is native for? 2.0? 2.1? 3.0? 3.2? However I wouldn't mind seeing some pictures if it is 3.2. Do graphics cards need to be designed for support of opengl 2.0, 3.0 etc like DX versions? Does the 5xxx series AMD/ATI cards support the newest opengl?

It is OpenGL 3.2. There is a AMD_vertex_shader_tessellator OpenGL extension avaialble on AMD cards since HD 2xxx: http://unigine.blogspot.com/2009/07/amd ... lator.html
Unigine supports this feature, but unfortunately it's completely broken in current video drivers from AMD. Their developers are aware of the problem and have already fixed it in their internal builds, but it will take 2 months more before these fixed drivers will be available publicly.

[skier]

i can't wait for DirectX 9 to be kicked out the door once and for all

then theres that DX10 guy, who knows whats up with him

[spitter]

Could you do a test using crossfire?

It takes ~ 4.5 minutes for the test to run and I benchmarked 4 cards at 4 resolutions/settings each 3 times to get an average of the scores.

That would be 48 runs at 4.5 minutes each or roughly 216 minutes Then I did the API runs, which is 12 more runs or 54 more minutes... I have 4.5 hours tied up in benchmark just these four cards... If I did CrossFire testing that is 4.5 more hours... I really just don't have the time right now.

This was just a for fun article and was posted on the weekend. If this article brings in big traffic and makes it worthwhile to run more tests I'll do a CrossFire update. Not trying to complain, but I always get the 'do more tests' and 'why didn't you do this' and people don't understand the amount of time involved to generate a couple charts. If it was automated it would be one thing, but this requires sitting in front of the screen for hours on end.

And I thought you did this for the parties, parts and wenches....

[Apoptosis]

I did 8 years ago when I was doing this for 100% complete fun as a hobby ;)

Not sure when it became a job, but it certainly has. I have seen that most sites close down after 3-5 years and that other site owners don't write much or are just now silent partners. Doing articles on benchmarks like this isn't for the hardware, the parties or for anything other than 'hey i want to see how this benchmark performs'... Call it my own desire to punish myself for a day...