View Full Version : Why would 2xAA give better Doom3 score than no AA with 66.93??

11-13-04, 12:23 PM
Decided to upgrade to 66.93 from 61.77 in anticipation of HL2.

I'm pleased that in the Dom3 Timedemo demo1 test I went from 53.3 running at 1600x1024 High Quality settings with no AA to 64.1 under 66.93. Nice in my book...

However... I've got an odd issue. When I enable 2x AA, my score jumps from 64.1 to 71.7 (very repeatable after running test 4 times). Going to 4x AA drops me to 52.4fps.

Why would RAISING AA to 2x make my score go UP and not down??



11-13-04, 12:44 PM
Doom 3 'High quality' setting forces 4x afaik, so setting 2x in the control panel will boost performance, regardless of your in-game setting.

11-13-04, 01:04 PM
Doom3 doesnt force AA afaik, it only has AF settings built in, like most recent games.

11-13-04, 01:18 PM
Well... I think I figured out my own problem.

It looks like it is related to a bug in the 66.93 drivers if you use Refresh Rate override. I was completely unaware of this issue (i.e. that refresh rate overide in the Nvidia Control Panel inflicts a MASSIVE hit to your framerates) until I stumbled across a thread here (http://www.nvnews.net/vbulletin/showthread.php?t=40365).

When I killed Nvida's Refresh Rate overide and replaced it with Refresh Force, my Doom 3 @ 1600x1024 High Quality No AA went from 64.1 to 86.6fps!!

That is one SERIOUS bug!

Personally, for people like myself who are "casual" enthusiasts, I think there should be a stickie just for this one as it is a HUGE performance hit.


11-13-04, 05:21 PM
I posted about this bug in the 66.93 feedback thread. I'm a newcommer to nVidia but I think that most people who use refresh rate override knew about this bug from the previous 66.xx drivers. It is shocking to know that this issue has been carried over to the latest drivers and that there is no mention of it in the release notes.

11-15-04, 11:52 AM
Excuse my ignorance... does the refresh rate bug affect the 65.xx driver family (I have 65.73)? I've long felt my Doom 3 performance is somewhat sub-par given my system (Athlon 3500+, 1GB RAM, 6800 Ultra - in fact wazooda has very similar components to me, mobo included) and I wonder if this could be the reason. I can bench 62-66fps in the timedemo at 16x12 HQ 8xAF, but find in-game gameplay is not nearly as smooth as I'd like - I've had to resort to playing with vsync disabled, as otherwise my fps regularly tank to below 20 (also fixedtic is tweaked to remove the framerate cap). Obviously image quality goes to sh*t due to tearing (refresh rate set to 72mHZ, as monitor only supports 75mHZ @ 16x12). Things will then get pretty choppy with any level of AA turned on.

However, given that my frame rates can bounce between 40fps and 200fps in-game, I'm not sure whether the refresh rate bug is the problem - especially as in the nvidia control panel I have refresh rate override set to 'let applications control refresh rate'. So probably not, but perhaps can someone can clarify things? Also, I take it RefreshForce is easy to install/apply?