HQ AF & Adaptive FSAA

Here we'll take a look of at the performance implications of using both High Quality Anisotropic Filtering and Adaptive FSAA simultaneously, using the same tests as before.

In this Far Cry test using both of these IQ features together, naturally, produces a greater performance hit, however at high resolution the performance reduction is greater than the sum of enabling the features individually. The increased  performance hit is likely due to the transparencies occurring on textures which are now being over sampled for FSAA purposes and 16x filtering is also being applied to them (some of which would have been reduced under angle dependant filtering).

The performance hit for enabling both features on out Half Life 2 test is much lower than that of the Far Cry test, which is again related to the lower frequency of transparencies here.

One of the focuses of R520 has been its image quality capabilities, and one area that ATI say that they have paid particular attention to is optimising 6x FSAA in order to reduce the performance impact and make it more usable in a wider variety of situations. We've seen throughout the testing that the R520 based boards are indeed more capable of rendering with FSAA enabled in relation to their predecessors, but by directly comparing the performance hit from 4x to 6x FSAA we can see if there have been further improvements for the efficiency of 6x FSAA specifically.

Under out Far Cry test we see that the performance difference from going from 4x FSAA to 6x is fairly slim for the X1800 XT, but this board continues to be mainly CPU bound by this test. Underlying that there are further optimisations are the differences in performance impacts between the X1800 XL and X800 XT as the XL has a significantly lower performance hit from going from 4x to 6x FSAA.

Half Life 2 displays a similar trend, although not quite as high in this case, in that generally speaking there is a lower performance penalty on the X1800's from going from 4x FSAA to 6x FSAA.