Originally Posted by Blacklash
If you disable DX10.1 the highest quality of transparency AA is not selectable. When left off you may select DX10 high quality transparency AA only. They seem to have decided to implement transparency AA as a DX10 or 10.1 choice. It works fine combined with DX11 Ultra quality lighting. If you want specifics, ask them. I wanted to test the highest quality of transparency AA available stacked on top of MSAA so I needed to have DX10.1 enabled.
If you want to discover this for yourself, download the bench and try it. You do not need the game to use it.
DX10 will be the highest AA option available for A tested objects if you disable 10.1. If you tick 10.1 the 10.1 AA option will appear under AA for A tested objects. I'd suggest you restart the bench when you change advanced settings to see if they saved correctly. If they did you will be able to see it when you open the advanced options area.
Again, my experience went like this; 1440x900 with 4xAA and 10.1 transparency AA all on or maxed = smooth with Crossfire X, 1920x1200 all on or maxed sans AA = smooth with Crossfire X, 1920x1200 with 4xAA and 10.1 transparency AA all on or maxed = frequent hard pauses that caused a subpar gaming experience.
I should note the CCC was left at driver default values. Things like "edge detect" were not used. I got the same behavior with my cards @ stock speed or overclocked.
Thanks for the reply.
I still don't believe there is a need to run 2 MSAA settings at the same time specially if there is such big performance hit. Did Google few benches and don't see other people using such.
Here is PCGH Stalker: Call of Pripyat - Benchmarks from DirectX 9 to DirectX 11
I'm sure they didn't run the DX10.1 & DX11 settings at the same time, since they show DX11 run faster.
from the picture gallery the image quality is better with DX11.
There is also long video included in the PCGH review and I don't notice any pauses there. PCGH is very much more than friendly NVIDIA site and if there was any problem they would show it fast.
Thanks for the download link for the bench but I already did run the bench in December last year and posted it here
I did run both DX10 and DX11 (not at the same time) and the DX11 was slightly faster. No problem with the minimum frames and both benches did run smooth.