- Joined
- Oct 27, 2008
I thought I'd share some of my results from my recent upgrade to the 4890 video card.
CPU: Q9550 @ 3.4 GHz
RAM: 2x 2GB DDR2 800
Chipset: X48
Card #1: SAPPHIRE TOXIC Radeon HD 4870 512MB 780/1000
Card #2: SAPPHIRE VAPOR-XL Radeon HD 4890 1GB 870/1050
Card #3: SAPPHIRE VAPOR-XL Radeon HD 4890 1GB 870/1050
So here they are in all their glory...
3DMark Vantage score for the 4870 flying solo
3DMark Vantage score for the 4890 flying solo
Now on to the Cross Fire
The 4870 and 4890's have different drivers to select from. This is to support the different ports on the 4890 like the native HDMI and also comes with an integrated physics portion that is supposed to increase physics performance.
I went with the current 4890 drivers and....
Success!
3DMark Vantage score for the 4870 & 4890 Cross Fire
3DMark Vantage score for the two 4890's Cross Fire
I kept the CPU and other hardware settings the same throughout testing to focus on the video cards alone. So the CPU scores will reflect 0% +/- 2% change which is really no change.
As you can see above there was an 11% performance gain over the 4870 with the new 4890.
The 4870/4890 Cross fire yielded significant video performance gains over the 4870 and 4890 alone of 70% and 53% respectively. The GPU Cloth and GPU Particles tests were actually unchanged or slower than the 4890 alone indicating the 4870 was holding the 4890 back. When the 4890’s were in CF the scores jumped back up and surpassed the single 4890.
As expected the difference between the two cross fire set ups is nearly the same as the difference between the single 4870 and 4890.
Two side notes…
1) I noticed some micro shuttering during the Parallax Occlusion Mapping (Complex Pixel Shader) test with the 2x 4890 CF that I did not observe with the 4870/4890 CF. This repeated it’s self and only on this test.
2) I accidently left the second 4890’s clock speeds the same as the 4870 and generated nearly identical results as the 4870/4890 CF in spite of the extra RAM being utilized
And if anyone is interested here is a description of the 3DMark Vantage tests
Overall, the differences are small from card to card and CF to CF so If you're on a budget, two 4870's or a combination of 4870 and 4890 is nearly the same performance for a lot less money. IMO
Let me know what you guys think...
CPU: Q9550 @ 3.4 GHz
RAM: 2x 2GB DDR2 800
Chipset: X48
Card #1: SAPPHIRE TOXIC Radeon HD 4870 512MB 780/1000
Card #2: SAPPHIRE VAPOR-XL Radeon HD 4890 1GB 870/1050
Card #3: SAPPHIRE VAPOR-XL Radeon HD 4890 1GB 870/1050
So here they are in all their glory...
3DMark Vantage score for the 4870 flying solo
3DMark Vantage score for the 4890 flying solo
Now on to the Cross Fire
The 4870 and 4890's have different drivers to select from. This is to support the different ports on the 4890 like the native HDMI and also comes with an integrated physics portion that is supposed to increase physics performance.
I went with the current 4890 drivers and....
Success!
3DMark Vantage score for the 4870 & 4890 Cross Fire
3DMark Vantage score for the two 4890's Cross Fire
I kept the CPU and other hardware settings the same throughout testing to focus on the video cards alone. So the CPU scores will reflect 0% +/- 2% change which is really no change.
As you can see above there was an 11% performance gain over the 4870 with the new 4890.
The 4870/4890 Cross fire yielded significant video performance gains over the 4870 and 4890 alone of 70% and 53% respectively. The GPU Cloth and GPU Particles tests were actually unchanged or slower than the 4890 alone indicating the 4870 was holding the 4890 back. When the 4890’s were in CF the scores jumped back up and surpassed the single 4890.
As expected the difference between the two cross fire set ups is nearly the same as the difference between the single 4870 and 4890.
Two side notes…
1) I noticed some micro shuttering during the Parallax Occlusion Mapping (Complex Pixel Shader) test with the 2x 4890 CF that I did not observe with the 4870/4890 CF. This repeated it’s self and only on this test.
2) I accidently left the second 4890’s clock speeds the same as the 4870 and generated nearly identical results as the 4870/4890 CF in spite of the extra RAM being utilized
And if anyone is interested here is a description of the 3DMark Vantage tests
Overall, the differences are small from card to card and CF to CF so If you're on a budget, two 4870's or a combination of 4870 and 4890 is nearly the same performance for a lot less money. IMO
Let me know what you guys think...
