Home  |  News  |  Reviews  | About Search :  HardWare.fr 

ProcessorsMotherboardsGraphics CardsMiscellaneousHard DisksSSD
Advertise on BeHardware.com
Review index:
Report: GeForce GTX 560 Ti vs Radeon HD 6950 1 GB
by Damien Triolet
Published on March 23, 2011

The new Catalysts
With the Catalyst 11.1a hotfix and as of the standard 11.2s, AMD has introduced two innovations.

The first concerns texture filtering and the associated optimisations. As you no doubt now know, a few months ago AMD took the controversial decision of reviewing the quality offered by default, so as to benefit from a modest gain in performance but with the result that there was additional flickering in textures.

AMD has revised its thinking and put into place a new compromise (Quality mode). This results bringing the Quality setting back up to what was the original default quality for the Radeon HD 5800s, although this time AMD has not disabled the optimisation which consists in limiting trilinear filtering to areas around mipmap level transitions. This is a worthwhile optimisation and one that NVIDIA also uses and which can be disabled by moving to HQ mode.
In spite of this, the GeForces filtering is still slightly cleaner on the GeForces, with less flickering. If this default setting is a problem for you, AMD also has an oddly named ‘Performance’ mode, which is identical to the Q setting but with a modification to LOD biases, reducing both texture sharpness and flickering.

We definitely approve of these developments which allow a return to the initial status quo. Of course the gain in performance linked to the new Quality setting, which we use here in our performance measurements, is lower than with previous versions, which themselves only give a 1 to 1.5% gain on average. At the end of the day, then, playing with the reputation of its products in terms of quality, only resulted in a minimum gain for AMD. Wouldn’t it have been better to deploy its resources elsewhere? To improve filtering on forthcoming GPUs for example?

AMD has taken the opportunity offered by these drivers to introduce a new tesselation optimisation. A new setting allows you to define the maximum level of tessellation allowed, fixed at 64 by Direct3D 11. The AMD Optimized setting automatically uses the application profile information implemented by AMD. As things stand, no tessellation factors have been limited in AMD profiles, but this does mean AMD is armed against any future games that push tessellation levels beyond useful levels. We did nevertheless disable this setting in our tests, going for ‘Use application settings’.

Note that you can also fix the maximum level of tessellation manually (no tessellation), 2, 4, 6, 8, 16, 32 and 64 (no limit). We also wanted to look at the impact of this setting in HAWX 2, which gives an enormous advantage to the GeForces, at 1920x1200 AA4x with a Radeon HD 6970:

Use application settings: 92 fps
Limited to 64: 92 fps
Limited to 32: 92 fps
Limited to 16: 93 fps
Limited to 8: 101 fps
Limited to 6: 101 fps
Limited to 4: 107 fps
Limited to 2: 108 fps
Limited to 1: 109 fps
Tessellation deactivated in the game: 130 fps

For comparison, here’s what we got with the GeForce GTX 570:

Tessellation activated in the game: 152 fps
Tessellation deactivated in the game: 199 fps

The test
The test
For this test, we used the same protocol as used in the Radeon HD 6900s test. The tests were carried out at 1920x1200, without FSAA, with MSAA 4x and with MSAA 8x, at maximum settings except in Metro 2033.

We made sure to test this mode on the GeForces, which isn’t always easy to work out. In the NVIDIA drivers, 8x antialiasing is in fact MSAA 4x with CSAA 8x which doesn’t give the same quality as MSAA 8x, which, for its part, is called 8xQ antialiasing. This is therefore what we tested.

Now one year after its release, we’ve decided that we can no longer treat DirectX 11 separately, especially as all the cards tested here are compatible with this API. DirectX 11, 10 and 9 games are therefore mixed together and we opted for very high settings even in the most demanding games. The most recent updates were installed and all the cards were tested with the most recently available drivers.

We decided to stop showing decimals in game performance results so as to make the graph more readable. We nevertheless note these values and use them when calculating the index. If you’re observant you’ll notice that the size of the bars also reflects this.

The Radeons and the GeForces were tested with the "quality" texture filtering setting. All the Radeons were tested with the Catalyst 11.1a hotfix (8.82.2) driver. All the GeForces were tested with beta 266.56 drivers.

Test configuration
Intel Core i7 975 (HT and Turbo desactivated)
Asus Rampage III Extreme
6 Go DDR3 1333 Corsair
Windows 7 64 bits
Forceware 266.56
Catalyst 11.1a hotfix

<< Previous page
Theoretical tests: geometry

Page index
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22
Next page >>
Need for Speed Shift  

Copyright © 1997- Hardware.fr SARL. All rights reserved.
Read our privacy guidelines.