[Tested] SAPPHIRE Radeon HD 6870 1024MB Review


SAPPHIRE Radeon HD 6870 Review Index


4 – SAPPHIRE Radeon HD 6870 Direct3D performances


4.1 Unigine Heaven (DirectX 11)

OpenGL tests have been finished on Unigine Heaven, what a better transition than starting Direct3D tests with Unigine Heaven 2.1, this time using the Direct3D 11 render path.

OpenGL 4 - Unigine Heaven 2.1



Settings: 1920×1080 fullscreen, tessellation: normal, shaders: high, 4X MSAA, 16X anisotropic filtering.

47.6 FPS, Scores: 1200 – ASUS ENGTX580
39.4 FPS, Scores: 991 – EVGA GeForce GTX 480
26.8 FPS, Scores: 674 – ATI Radeon HD 5870
25.9 FPS, Scores: 653 – ASUS EAH6870
25.3 FPS, Scores: 637 – SAPPHIRE HD6870
25.6 FPS, Scores: 646 – MSI N460GTX Cyclone 768D5
16.6 FPS, Scores: 419 – MSI R5770 Hawk


4.2 MSI Kombustor (Direct3D 9)

MSI Kombustor 1.1.3 has been used for the test. MSI Kombsutor is a Direct3D 9, 10, 11 and OpenGL 2 and 3 stress test and benchmark. I only tested the Direct3D 9 rendering.

MSI Kombustor, Direct3D 9

Settings: 1920×1080 fullscreen, no AA, no postFX, 60sec, Unlock power draw CHECKED (the real benchmark mode!).

2378 (40FPS) – ASUS ENGTX580 *** OCP UNLOCKED ***
1991 – EVGA GeForce GTX 480
1962 – ATI Radeon HD 5870
1833 – ASUS EAH6870
1816 – Sapphire Radeon HD 6870
1738 – ASUS ENGTX580
1595 – MSI GeForce GTX 470
1201 – MSI N460GTX Cyclone 768D5 OC
943 – MSI R5770 Hawk (Cat 10.7)


4.3 NVIDIA Island Demo: Tessellation (Direct3D 11)

NVIDIA Island demo is a D3D11 demo focused on, what a surprise, tessellation!

NVIDIA Island DX11 demo



Settings: windowed (default size) and default params (tess factor: 12).

FPS: 57 – ASUS ENGTX580
FPS: 46 – EVGA GTX 480
FPS: 25 – MSI N460GTX Cyclone 768D5 OC
FPS: 13 – ASUS EAH6870
FPS: 13 – Sapphire Radeon HD 6870
FPS: 11 – ATI Radeon HD 5870
FPS: 10 – MSI R5770 Hawk


4.4 DirectX SDK SubD11: Tessellation (Direct3D 11)

SubD11 is a Direct3D tessellation demo from the DirectX 11 SDK June 2010.

SubD11 DX11 demo

Settings: 1920×1080 windowed, tessellation: 8X

FPS: 250 – ASUS ENGTX580
FPS: 220 – EVGA GTX 480
FPS: 135 – MSI N460GTX Cyclone 768D5 OC
FPS: 74 – ASUS EAH6870
FPS: 73 – Sapphire Radeon HD 6870
FPS: 60 – ATI Radeon HD 5870
FPS: 53 – MSI R5770 Hawk

Settings: 1920×1080 windowed, tessellation: 16X

FPS: 118 – ASUS ENGTX580
FPS: 102 – EVGA GTX 480
FPS: 59 – MSI N460GTX Cyclone 768D5 OC
FPS: 19 – ASUS EAH6870
FPS: 18 – Sapphire Radeon HD 6870
FPS: 16 – ATI Radeon HD 5870
FPS: 15 – MSI R5770 Hawk

Settings: 1920×1080 windowed, tessellation: 24X

FPS: 51 – ASUS ENGTX580
FPS: 44 – EVGA GTX 480
FPS: 27 – MSI N460GTX Cyclone 768D5 OC
FPS: 8 – ASUS EAH6870
FPS: 8 – Sapphire Radeon HD 6870
FPS: 8 – ATI Radeon HD 5870
FPS: 7 – MSI R5770 Hawk

Settings: 1920×1080 windowed, tessellation: 31X

FPS: 30 – ASUS ENGTX580
FPS: 26 – EVGA GTX 480
FPS: 16 – MSI N460GTX Cyclone 768D5 OC
FPS: 5 – ASUS EAH6870
FPS: 5 – Sapphire Radeon HD 6870
FPS: 4 – ATI Radeon HD 5870
FPS: 4 – MSI R5770 Hawk


4.5 3DMark06 (Direct3D 9)

3DMark06 is one of the most used Direct3D 9 benchmark. It’s still interesting to have the scores for this benchmark.

Settings: default settings.

Score: 22379 (SM2: 8212, SM3: 11585, CPU: 5805) – ASUS ENGTX580
Score: 22162 (SM2: 8296, SM3: 11321, CPU: 5737) – EVGA GTX480
Score: 21397 (SM2: 8377, SM3: 10419, CPU: 5683) – Radeon HD 5870
Score: 21124 (SM2: 8337, SM3: 10146, CPU: 5689) – SAPPHIRE Radeon HD 6870
Score: 20484 (SM2: 8199, SM3: 9470, CPU: 5806) – MSI N460GTX Cyclone

Remark: this test is no longer adapted for modern GPUs. All cards have similar scores, meaning the benchmark is CPU-bound…


4.6 Aquamark 3 (Direct3D 9)

Aquamark 3 is a popular Direct3D 9 benchmark. Under Windows Vista or Seven, you have to install this patch. I posted the scores just for the fun because I don’t know if this tool is still representative of modern cards like GeForce GTX 500 / Radeon HD 6000…

Settings: default: 1024×768, fullscreen, 0XMSAA, 4X aniso.

163.7 FPS (GFX:28221, CPU:19477) – Radeon HD 5870
161.18 FPS (GFX:27513, CPU:19459) – Sapphire Radeon HD 6870
154.60 FPS (GFX:24018, CPU:21689) – MSI N460GTX Cyclone
153.74 FPS (GFX:23931, CPU:21511) – GeForce GTX 480

Remark: like 3DMark06, this test is no longer adapted for modern GPUs. All cards have similar scores, meaning the benchmark is CPU-bound…

SAPPHIRE Radeon HD 6870 Review Index


13 thoughts on “[Tested] SAPPHIRE Radeon HD 6870 1024MB Review”

  1. JeGX Post Author

    GTX 580 Crysis score is for the next review. I’m testing a GTX 580 from EVGA…

  2. Lokavidu

    guys seriously consider putting the index on the bottom of the site too.. using home key is a bit annoying and unfamiliar

  3. iPristy

    OK, will you then compare 6870 vs 580 ?

    Radeons are fast enough in games, artificial bench demos it’s about the same as Sunspider bench who cares about that.

  4. JeGX Post Author

    @iPristy: sure, all previous scores (HD 5870, HD 6870, GTX 4xx) will be in EVGA GTX 580 review.

  5. Pingback: [Tested and Burned] EVGA GeForce GTX 580 SC Review - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  6. Pingback: [GPU Tool] ShaderToyMark 0.2.0 Demomakers’ Benchmark Available with Better SLI and CrossFire Support - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  7. Pingback: AMD Catalyst 10.12 WHQL (OpenGL 4.1, OpenCL 1.1) - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  8. Pingback: [Tested and Burned] SAPPHIRE Radeon HD 6970 2048MB Review - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  9. Pingback: [Tested] ASUS Radeon HD 6950 2GB GDDR5 Review - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  10. Pingback: [Tested and Burned] ASUS GeForce GTX 560 Ti DirectCU II TOP Review - 3D Tech News, Pixel Hacking, Data Visualization and 3D Programming - Geeks3D.com

  11. kromar

    these opengl tests do not at all reflect the performance of real opengl applications.
    when using the latest fermi architecutre in opengl heavy 3d apps like blender, maya and so on the true weakness of these cards emerge.

    it would be really great if you could add a test done in blender to the ones yu already did to see how the “real” opengl performance is.

Comments are closed.