Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

did you test dx11 with heaven?

drbaltazardrbaltazar Member UncommonPosts: 7,856

i tried heaven benchmark 1.0 is the engine a ressource hog or am i set-up wrong

i tried heaven a x4 everything on at 1650 x1050  men i only get about 15 fps avergage

what is a good value for this test

ty for info if any maybe i activated stuf that werent supposed to be active ?

 

ati 5770 1gig idx11 is my card

amd x2 550 be

4 gig of 1600 ram

but it does look way better then what i saw in youtube me its cg quality look but its not a cg

grin

Comments

  • noquarternoquarter Member Posts: 1,170

    That's about what I get on my 5770. Also close to what this review shows http://benchmarkreviews.com/index.php?option=com_content&task=view&id=408&Itemid=72&limit=1&limitstart=15

    It's not a great representation of an actual game since benchmarks are often more demanding than real games but I bought my 5770 knowing it might need a 2nd 5770 in crossfire later on or to replace with a 5850 to really handle DX11. We'll see when Aliens vs Predator comes out. The 5000 series is suppose to crossfire really well though.

  • CleffyCleffy Member RarePosts: 6,412

    With tesselation, you can get all cards that use a tesselator to run at exactly 60 fps.  Some cards will have more polies, but its a controllable value.  15 fps is actually pretty good for your setup.  Most people running the benchmark with a fresh top of the line system running your graphics card gets around 22 fps.  I would not put that in the realm of realism for a system that has been used for several weeks or longer.  Naturally the HD5870 gets about twice your fps.

    I really don't like it when people completely misunderstand what DX11 does.  Like the guy above me saying DX11 causes some type of performance hiccup.  The thing thats exciting about DX11 is you don't have to work on games like was done in the past.  You can make a basic model that runs at 60 fps on an IGP, then displacement map it for higher end video cards.  Best of all you can just work on DX11, and support all the way down to DX9 with no additional work.  There isn't some mystical DX11 wall that makes some peoples hardware choices idiotic.  The developer has easy automatic control to give the player the best visual experience regardless of hardware if they put the effort into it.

    Things like crossfire depend nearly entirely on the driver support and internal hardware, or the engine.  Either the driver developer has to go out of their way to make crossfire work on a particular game, or the developer has to make crossfire work on their engine.  You should not see any difference in the effect of crossfire between an HD5870 and an x800 Pro, unless AMD designed a better crossfire processor.  Something along the lines of a Lucid Hydra that seperates DX calls between different GPUs, composites them, then renders for a near 100% increase in performance.

    I'm really not a fan of the current crop of next-gen engines.  They just don't do it for feature support.  Unreal Engine has gotten harder to modify, Crytek lacks the necessary next-gen features along with Unreal but is easier to create different gameplay, and the Unity Engine is underpowered.  Pretty much the best middle-ware engine has become the Ogre Engine.  It doesn't do much except sets up the basic interconnects.  Because it does practically nothing, its easier to add on what you want ontop without the engine trying to stop you.

  • noquarternoquarter Member Posts: 1,170

    Sorry but right now there IS a performance hiccup for using tessellation. Has nothing to do with the technology or the possibilities of it, but the way it's implemented in the first couple generations of DX11 games this will be true.


    For the next few years we are still going to be seeing a high quality base model with tessellation enhancements the low end may not be able to enable due to performance hit since tessellated polys still need to go through all the shader processes.


    I haven't seen any engine yet that will be doing dynamic tessellation in the sense that the degree of tessellation changes based on how many polys your GPU can handle (which is ultimately the best part), I've only seen engines that tessellate based on distance. Which means with tessellation on you get a certain # of polys beyond the base model forced into your GPU that it may not be able to handle.


    I think once the install base is entirely DX11 we'll start seeing engines that use medium quality base models with heavy dependence on dynamic tessellation to always provide as many polys as the GPU can handle. At that point, any DX11 card will handle DX11 properly and it will just be faster card = more polys.


    Until then we will have games that use distance based tessellation on high quality base models that require a faster DX11 card just to turn on, since the current implementation is "Here's 100,000 polys + 100,000 more since you have tessellation!" instead of "Here's anywhere between 50,000 to 200,000 polys since you have tessellation!"

  • coffee2coffee2 Member Posts: 10

    I got about 23FPS in 1400x900 on my 5770 Viper-X.  Which is what I expected.

    I guess the top end 5870 which costs almost 3 times as much as the 5770 (which is just a mid range card) would be closer to 50FPS.

    AvP2 is out in a few weeks that features DX11 tesilation so looking forward to testing it out.

    AKA "coffee" - serving a temporary ban cus I made a Troll cry.

  • CleffyCleffy Member RarePosts: 6,412

    When you look at the entire market, there is a clear benefit in approaching tesselation as I outlined it.  Most computers use an IGP, of those with a discrete graphics card atleast 30% have a tesselator on them as ATI has included it for generations.  Outside of the PC the XBox360 also has a tesselator.  Its not like I am talking about a small market base for this approach.  Its a much bigger base then you would have going for today's standard 1 million triangle scene.  Also Tesselation is 4x polies at each step.

  • LeononaLeonona Member UncommonPosts: 225

    I my opinion the main problem is that the tesselator of the ati 5000-series is just rather poor. The architecture was never meant to handle dx11 and as a result performance is as you see. Nvidias fermi on the other hand was made for dx11 and as a result can handle a lot more geometry. Performane advantage for fermi over ati hd5000 will not be very great in dx 9 and dx10 but in dx11 it will be significant.

  • CleffyCleffy Member RarePosts: 6,412

    I disagree with that as well.  When you look at the actual architecture of the AMD HD line, it was purpose built for displaying a large amount of polies.  This is why when you can't afford a workstation card for 3D work, you can alternately use an AMD Desktop card.  It renders polies 3 times more effeciently then its nVidia counterparts due to the structure of AMDs stream processor.  As a result AMD cards for the last few generations have been slightly worse at swapping and using textures, but astronomically better at rendering polies and doing calculations.  Its doubtful that Microsoft developed DX11 Tesselation support off of any other architecture aside from AMD as AMD is the only one who has a working Tesselator.

  • LeononaLeonona Member UncommonPosts: 225

    Well...it's possible that ati built their architecture to handle 'many' polys and that it can handle more than nvidias current cards, but when you compare hd5000 to what fermi will be coming with, the difference is vast in favor of fermi. Nvidia built what they call the polymorph engine for their gf100-chip and it really leaves ati in the mud. It can handle a LOT more. Try having a look at this page:

    http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/27892-nvidia-s-geforce-gf100-under-microscope-7.html

  • rygar218rygar218 Member UncommonPosts: 332

    Anyone know when DX11 is to offically release?

  • SomnulusSomnulus Member Posts: 354

    All right, perhaps I have some kind of magical rig. I honestly don't know.

     

    Here are the results for my unigine Heaven benchmarks. I did one benchmark in full screen and one windowed. FPS and scores are enlarged for easier viewing. Both of these are plain text cut and paste from the saved html files from both benchmarks.

     

    Full Screen:

     

    Unigine Heaven Demo v1.0

    FPS: 57.4

    Scores: 1446

    Hardware Binary: Windows 32bit Visual C++ 1500 Release Oct 22 2009

    Operating system: Windows 7 (build 7600) 64bit

    CPU model: Intel(R) Core(TM)2 Extreme CPU X9650 @ 3.00GHz

    CPU flags: 3064MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 HTT

    GPU model: ATI Radeon HD 4800 Series 8.682.2.0 512Mb

    Settings Render: direct3d11

    Mode: 1680x1050 fullscreen

    Shaders: high

    Textures: high

    Filter: trilinear

    Anisotropy: 16x

    Occlusion: enabled

    Refraction: enabled

    Volumetric:enabled

    Unigine Corp. © 2005-2009

     

    Windowed:

     

    Unigine Heaven Demo v1.0

    FPS: 33.8

    Scores: 852

    Hardware Binary: Windows 32bit Visual C++ 1500 Release Oct 22 2009

    Operating system: Windows 7 (build 7600) 64bit

    CPU model: Intel(R) Core(TM)2 Extreme CPU X9650 @ 3.00GHz

    CPU flags: 3064MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 HTT

    GPU model: ATI Radeon HD 4800 Series 8.682.2.0 512Mb

    Settings Render: direct3d11

    Mode: 1680x1050 windowed

    Shaders: high

    Textures: high

    Filter: trilinear

    Anisotropy: 16x

    Occlusion: enabled

    Refraction: enabled

    Volumetric: enabled

    Unigine Corp. © 2005-2009

     

    Tesselation was enabled, Vsync was enabled, and I had anisotropy set to 16x. I did not have anti-aliasing on, because for whatever reason ATI tends to not like running anisotropic filtering and anti-aliasing in the engine software at the same time. I normally set the level of anti-aliasing in the Catalyst driver.

     

    Also, I am running dual Sapphire HD4870s in Crossfire. My system RAM is PC8500 4GB (OCZ Reaper).

     

    So nothing magical there at all, but my FPS seems to be outstanding compared with what others here are reporting.

     

    Overall, it looked good, although for a benchmark there really wasn't much going on and the models themselves seemed to be rather low poly. I would be interested to see how it handles some higher-poly models and some serious FPS-draining particle effects.

     

    Abbatoir / Abbatoir Cinq
    Adnihilo
    Beorn Judge's Edge
    Somnulus
    Perfect Black
    ----------------------
    Asheron's Call / Asheron's Call 2
    Everquest / Everquest 2
    Anarchy Online
    Shadowbane
    Dark Age of Camelot
    Star Wars Galaxies
    Matrix Online
    World of Warcraft
    Guild Wars
    City of Heroes

  • LeononaLeonona Member UncommonPosts: 225

    @ Somnulus

    There's a good reason you get so good performance and the geometry are low poly. Your card is a hd4800 and it does not support dx11.

  • CleffyCleffy Member RarePosts: 6,412

    DX11 is backwards compatible down to DX9.  This means for a DX10.1 card like the HD4870 with a tesselator, its possible to have the tesselator on as you can tell from the dip he had in FPS when he turned it on.  Also it would be obvious there is tesselation when you view the benchmark.  The HD4870 has a 256-bit bus which allows it to perform better then the HD5770 with its 128-bit bus.  Also you are using 2 cards not 1 like the above poster and the results from other cards.

  • noquarternoquarter Member Posts: 1,170


    Originally posted by Cleffy
    DX11 is backwards compatible down to DX9.  This means for a DX10.1 card like the HD4870 with a tesselator, its possible to have the tesselator on as you can tell from the dip he had in FPS when he turned it on.  Also it would be obvious there is tesselation when you view the benchmark.  The HD4870 has a 256-bit bus which allows it to perform better then the HD5770 with its 128-bit bus.  Also you are using 2 cards not 1 like the above poster and the results from other cards.

    The DX11 tessellator is a different implementation than the tessellator in the pre-5000 series Radeons, so the last gen still won't tessellate DX11 games unless a game is specifically designed to use the old hardware implementation and not just the DX11 tessellator and I don't know anything that use the old tessellator.


    So even though it says DX11 there's no tessellation.. easy to check too by just turning on wireframe or looking at the roofing or cobblestones.


    The 2 results he posted are full screen and windowed. As far as I know Crossfire doesn't work in Windowed mode which explains there's a large FPS difference between the 2.

  • drbaltazardrbaltazar Member UncommonPosts: 7,856

    dx11 isnt cool because of tessalation its the amount of data parallee it can process instead of just using 1 pipeline

    true it wont mather for now since we go two other problem limiting the capability of those

    32 bit game and no microsoft donnybrook.wich im sorry to say will be needed to take full advantage of dx11 feature

    i ignore tessalation for the next 3 year since game make already cant keep up with the amount of data they have to pipeline

    to game so if they added tessalation it would make game worst

    but dx11 card have lot of other hidden feature that will make them very nice to have

  • drbaltazardrbaltazar Member UncommonPosts: 7,856
    Originally posted by Somnulus


    All right, perhaps I have some kind of magical rig. I honestly don't know.


     
    Here are the results for my unigine Heaven benchmarks. I did one benchmark in full screen and one windowed. FPS and scores are enlarged for easier viewing. Both of these are plain text cut and paste from the saved html files from both benchmarks.


     
    Full Screen:


     
    Unigine Heaven Demo v1.0

    FPS: 57.4

    Scores: 1446

    Hardware Binary: Windows 32bit Visual C++ 1500 Release Oct 22 2009

    Operating system: Windows 7 (build 7600) 64bit

    CPU model: Intel(R) Core(TM)2 Extreme CPU X9650 @ 3.00GHz

    CPU flags: 3064MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 HTT

    GPU model: ATI Radeon HD 4800 Series 8.682.2.0 512Mb

    Settings Render: direct3d11

    Mode: 1680x1050 fullscreen

    Shaders: high

    Textures: high

    Filter: trilinear

    Anisotropy: 16x

    Occlusion: enabled

    Refraction: enabled

    Volumetric:enabled

    Unigine Corp. © 2005-2009

     
    Windowed:


     
    Unigine Heaven Demo v1.0

    FPS: 33.8

    Scores: 852

    Hardware Binary: Windows 32bit Visual C++ 1500 Release Oct 22 2009

    Operating system: Windows 7 (build 7600) 64bit

    CPU model: Intel(R) Core(TM)2 Extreme CPU X9650 @ 3.00GHz

    CPU flags: 3064MHz MMX SSE SSE2 SSE3 SSSE3 SSE41 HTT

    GPU model: ATI Radeon HD 4800 Series 8.682.2.0 512Mb

    Settings Render: direct3d11

    Mode: 1680x1050 windowed

    Shaders: high

    Textures: high

    Filter: trilinear

    Anisotropy: 16x

    Occlusion: enabled

    Refraction: enabled

    Volumetric: enabled

    Unigine Corp. © 2005-2009

     
    Tesselation was enabled, Vsync was enabled, and I had anisotropy set to 16x. I did not have anti-aliasing on, because for whatever reason ATI tends to not like running anisotropic filtering and anti-aliasing in the engine software at the same time. I normally set the level of anti-aliasing in the Catalyst driver.
     
    Also, I am running dual Sapphire HD4870s in Crossfire. My system RAM is PC8500 4GB (OCZ Reaper).


     
    So nothing magical there at all, but my FPS seems to be outstanding compared with what others here are reporting.
     
    Overall, it looked good, although for a benchmark there really wasn't much going on and the models themselves seemed to be rather low poly. I would be interested to see how it handles some higher-poly models and some serious FPS-draining particle effects.
     

     

    your graphic card isnt dx11 .it goes only as high as dx10.1 .so this means no tessalation if thats what you were reffering to

    only 5xxx serie as the physical thing needed for dx11 full feature!

  • CleffyCleffy Member RarePosts: 6,412

    Adding tesselation is barely any additional work.  When it comes to an asset its keeping a displacement map instead of trashing it.  As far as coding, AMD has already coded the shader.

  • SomnulusSomnulus Member Posts: 354
    Originally posted by Leonona


    @ Somnulus
    There's a good reason you get so good performance and the geometry are low poly. Your card is a hd4800 and it does not support dx11.

     

    That makes perfect sense. Thanks for the information, folks.... I knew there wasn't some kind of twist in the fabric of the universe that was allowing me to get better performance on the same tests.

    Ah well... I was going to upgrade to dual HD5870s 1 GB anyway.

    Abbatoir / Abbatoir Cinq
    Adnihilo
    Beorn Judge's Edge
    Somnulus
    Perfect Black
    ----------------------
    Asheron's Call / Asheron's Call 2
    Everquest / Everquest 2
    Anarchy Online
    Shadowbane
    Dark Age of Camelot
    Star Wars Galaxies
    Matrix Online
    World of Warcraft
    Guild Wars
    City of Heroes

  • CleffyCleffy Member RarePosts: 6,412

    Ah, that sucks.  I thought for sure they would implement tesselation on older cards that had the hardware.  I thought DX11 tesselation would just be an easier to access version of DX10 / 9.0c instanced based tesselation.

  • noquarternoquarter Member Posts: 1,170


    Originally posted by Cleffy
    Ah, that sucks.  I thought for sure they would implement tesselation on older cards that had the hardware.  I thought DX11 tesselation would just be an easier to access version of DX10 / 9.0c instanced based tesselation.

    Yea I know, especially since AMD told Microsoft how they wanted tessellation to work in DX11 lol

Sign In or Register to comment.