[Test] OpenGL 4.0 and Direct3D 11 Tessellation: GTX 480 vs GTX 470 vs HD 5870 vs HD 5770


OpenGL 4.0 tessellation - GTX 480 vs HD 5870



Here is an update of this post about Unigine Heaven 2.1 scores in OpenGL 4.0 and Direct3D 11 with tessellation in extreme mode.

Now that the Catalyst 10.5 driver is out with OpenGL 4.0 support, we can complete the test and see how Cypress transistors can play with OpenGL 4.0 tessellation.

First thing, Unigine Heaven works fine with Radeon HD 5000 series with OpenGL 4.0. So Catalyst 10.5 brings bugfixes compared to Catalyst 10.4 with GL 4.0 preview.

Here is an overview of the scores (extreme tessellation in 1920×1080):

– GeForce GTX 480: 916 (OpenGL 4.0)
– GeForce GTX 480: 970 (Direct3D 11)

– GeForce GTX 470: 719 (OpenGL 4.0)
– GeForce GTX 470: 768 (Direct3D 11)

– Radeon HD 5870: 316 (OpenGL 4.0)
– Radeon HD 5870: 532 (Direct3D 11)

– Radeon HD 5770: 258 (OpenGL 4.0)
– Radeon HD 5770: 364 (Direct3D 11)
– Radeon HD 5770 Crossfire: 695 (Direct3D 11) – OpenGL CrossFire does not work :(



Both NVIDIA and AMD offer better performance in Direct3D 11 than in OpenGL 4.0. But in the case of AMD, it seems the OpenGL 4.0 driver has really poor performances compared to the D3D one. But I’m sure that AMD’s OpenGL team will improve OpenGL tessellation speed.

Fun fact: according to this test, the GTX 480 is around 80% faster than the HD 5870 in Direct3D 11 and is around… 190% faster in OpenGL 4.0.. Please let me know if you notice an error in my numbers… It’s just not possible!



Here are the details of the test:

Testbed:
– Intel Core2 Extreme X9650 @ 3GHz
– 4GB DDR3
– Mobo: MSI P45D3 Platinium
– PSU: Antec TruePower Quttro 850W
– Win7 64-bit
Unigine Heaven 2.1



ATI Radeon HD 5870

ATI Radeon HD 5870 Scores

– R5770 clocks – GPU:850MHz and memory:1200MHz
Catalyst 10.5

OpenGL 4.0
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 316
– FPS: 12.6
– Min FPS = 2.9
– Max FPS = 70.3

Direct3D 11
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 532
– FPS: 21.1
– Min FPS = 6.6
– Max FPS = 74.6



MSI R5770 Hawk

MSI R5770 Scores

– R5770 clocks – GPU:875MHz and memory:1200MHz
Catalyst 10.5

As usual, CrossFire is not supported in OpenGL app (no Xfire profile…) :(

OpenGL 4.0 – Single HD 5770
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 258
– FPS: 10.3
– Min FPS = 3.8
– Max FPS = 43.6

Direct3D 11 – Single HD 5770
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 364
– FPS: 14.4
– Min FPS = 5.9
– Max FPS = 45.0

On the other hand, in Direct3D 11, Crossfire is nicely supported:
Direct3D 11 – CrossFire of MSI’s R5770 Hawk
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 695
– FPS: 27.6
– Min FPS = 8.9
– Max FPS = 89.6



EVGA GTX 480

EVGA GTX 480 Scores

– GTX 480 clocks – GPU:700MHz, memory:1848MHz and shader:1401MHz
– Display drivers: R257.15

OpenGL 4.0
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 916
– FPS = 36.4
– Min FPS = 6.1
– Max FPS = 89.4

Direct3D 11
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 970
– FPS = 38.5
– Min FPS = 17.6
– Max FPS = 95.5


GeForce GTX 470 Scores

– GTX 470 clocks – GPU:607MHz, memory:1674MHz and shader:1215MHz
– Display drivers: R257.15

OpenGL 4.0
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 719
– FPS = 28.5
– Min FPS = 5.9
– Max FPS = 70.4

Direct3D 11
Res: 1920×1080 fullscreen
Tessellation mode: Extreme
Score: 768
– FPS = 30.5
– Min FPS = 13.5
– Max FPS = 76.1


23 thoughts on “[Test] OpenGL 4.0 and Direct3D 11 Tessellation: GTX 480 vs GTX 470 vs HD 5870 vs HD 5770”

  1. Leith Bade

    Don’t forget NVIDIA paid/bribed/whatever Unigine to add ‘Extreme’ tesselation in Heaven 2.0

    They did this when they launched the 480 so they could release benchmarks to show the 480’s “superior” tessellator over the 5870.

    Rerun your benchmarks with “Normal” tesselation and see if ATI gets a lot closer to NVIDIA.

  2. JeGX Post Author

    I will test again with normal mode when Catalyst 10.6 will be published. It takes so much time to benchmark several cards in GL4 and D3D with Unigine heaven.

  3. Leith Bade

    Yeah lol.

    It took me a while to get D3D and OGL for my 5850. You have to run through benchmark twice so that it loads all the textures etc from HDD and wont lag.

  4. Fool Gamer

    depend on driver their release_!
    my oppinion is
    ati hardware is better than nvidia
    like (low TDP),(first vga support DX11)
    nvidia take longtime to release their DX11 vga
    ok
    just that

    COol

  5. Fool Gamer

    oh i forget ask you ?
    how about DX9c perfomance with lastest driver?
    thank

  6. JeGX Post Author

    @Fool Gamer: try to change your email address because your comments are still in the spambox…

  7. rocco

    nice result for 5770 crossx better then 5870 ….when i back home i try my shappire crossfire 5770 if have the same result

  8. Pingback: ATI Catalyst 10.5 With Official OpenGL 4.0 Support and 18 New Extensions - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  9. Sceptical

    There is just too much overlap with Unigine and Nvidia. Didn’t Nvidia buy Unigine 1.0 and users been reporting similar scores and performance in Unigine 2.0 as 1.0 now? It’s way too fishy.

    Why not try using something like Stone Giant instead?

  10. applejack

    ^ also with 2xAA according to that table,
    and that is with 197.75 WHQL, not the latest beta driver…

  11. Yno

    Hi,

    Is there any bench under Linux for Heaven? Would be interesting I guess :]

    By the way, do you read emails sent to tips at geeks3d dot com ? I sent one a month ago but never got any response so far.

    Thanks!

  12. Pingback: [TEST] AvP DX11 Tessellation Battle: GTX 480 vs GTX 470 vs HD 5870 vs HD 5770 - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  13. Pingback: [INFO] Aliens vs. Predator DirectX 11 Benchmark - Page 6 - CHIP Forum

  14. Pingback: [INFO] Aliens vs. Predator DirectX 11 Benchmark - Page 6 - CHIP Forum

  15. ca$per

    I remember we’ve ran unigine tropics throught gDebugger to see what it does in OpenGL rendering mode. And the results were…… strange. Unigine was doing some very useless work, like binding textures with 0x0 or 1×1 size, switching vertex buffers one after another without rendering, changing render states, also without rendering. But that was one-two years ago. Perhaps now everything works better.
    Can anyone check that? gDebugger is available as a free trial for 7 days. It’s just i’m not able to do that right now.

  16. Pingback: Primi rumors su un'ipotetica GTX 490 - [NEWS] - Pagina 4 | Forum -X- NextHardware

  17. spooky

    just been doing some tests and the opengl difference with nvidia atleast seems negligable now with the latest beta drivers (260.63)

    cpu: core2duo e6600 @ 2.4
    mem: 8gig ddr2
    gfx: msi gtx 470 @ core 766, shader 1532, mem 1674

    Unigine Heaven 2.1 @ 1920×1080, extreme tesselation

    opengl score -880
    DX11 score – 888

Comments are closed.