PDA

View Full Version : opengl AF Optimizations not disabling in HQ mode


Pages : [1] 2

rwolf
03-03-06, 01:23 AM
http://www.beyond3d.com/forum/showthread.php?p=709375&posted=1#post709375

It appears that opengl benchmarks are scewed because AF optimizations are not turned off in HQ mode.

OWA
03-03-06, 02:26 AM
It looks like they might have some sort of profile bug but for me, it didn't make any difference. Just setting the HQ mode I got 107.8 fps in Doom 3 timedemo. Disabling the optimizations individually as well, I still got 107.8 fps.

Also, even if there was a bug in the profiles, most sites seem to benchmark Cat AI on vs Quality which means optimizations are on for ATI and Nvidia. So, if there was indeed a bug, it would screw up a few of the results. I don't think there is one though but I've only tried Doom 3 so far.

Edit: I guess I need to read the source article if there is a translation. They must be doing something else differently as well. I just tried Pacific Fighters, which is very dependent on AF, and got:

avg: 57.62, Min: 29, Max: 63 with just HQ checked
avg: 57.69, Min: 29, Max: 63 with HQ and all opts individually disabled as well

Edit 2: I wonder if it could be some global setting issue. Since I keep my global settings on High Quality as well, that might be why I don't see any difference. I guess I'll try setting that to Quality to see what happens.

Edit 3: Riddick Results:

Avg: 98.39, Min: 4.25, Max: 291.04 just HQ checked
Avg: 99.36, Min: 12.80, Max: 278.13 HQ checked, all opts individually disabled as well

tertsi
03-03-06, 02:54 AM
Yeah, this is an old bug in NVIDIA control panel (affects only OpenGL apps) which I've found long time ago, like year or so (Feb 2005).

I did some tests with 7800GTX256 (+ IQ checked before tests with AF-tester) :

Doom 3
----------
16x12 + all maxed (no 3mood tweaks) + timedemo demo1

AF 16x + quality + all optimizations 91fps
AF 16x + quality - all optimizations 87fps
AF 16x + high quality + all optimizations 89fps
AF 16x + high quality - all optimizations 87fps

:: Overall: 3-4% performance hit.

Quake 4
----------
16x12 + all maxed (no Quak4 tweaks) + timedemo testi1 (singleplayer demo)

AF 16x + quality + all optimizations 78fps
AF 16x + quality - all optimizations 74fps
AF 16x + high quality + all optimizations 74fps
AF 16x + high quality - all optimizations 69fps

:: Overall: 5-11% performance hit.

16x12 + all maxed (no Quak4 tweaks) + timedemo testi2 (multiplayer demo)

AF 16x + quality + all optimizations 122fps
AF 16x + quality - all optimizations 115fps
AF 16x + high quality + all optimizations 117fps
AF 16x + high quality - all optimizations 110fps

:: Overall: 4-10% performance hit.

CoR
----------
12x10 + all maxed + timedemo testi1 (custom demo)

AF 16x + quality + all optimizations 98fps
AF 16x + quality - all optimizations 94fps
AF 16x + high quality + all optimizations 95fps
AF 16x + high quality - all optimizations 91fps

:: Overall: 3-7% performance hit.

OWA
03-03-06, 03:14 AM
I'm not seeing that. Could I be CPU limited making my results come out the same each time? I've tried Doom 3, CoR, and Pacific Fighters and get the same results no matter what. I'm only varying by like .1, .2, etc. fps. In fact, my riddick score went up slightly with the HQ - all opts. So, I must be doing something wrong.

RejZoR
03-03-06, 04:32 AM
Well i get killer quality in game Rune with 8xS and 16xAF (High Quality preset).
I'm running the game through customized OpenGL engine made by Chris Dohnal.

Kamel
03-04-06, 07:48 AM
I'm not seeing that. Could I be CPU limited making my results come out the same each time? I've tried Doom 3, CoR, and Pacific Fighters and get the same results no matter what. I'm only varying by like .1, .2, etc. fps. In fact, my riddick score went up slightly with the HQ - all opts. So, I must be doing something wrong.

crank up the resolution and anti aliasing and find out.

ginfest
03-04-06, 07:58 AM
Wow rwolf thanks for taking the time to come here and inform us. Now run back to B3D and arouse the ATI crew there with tales of your adventures here, lol. Oh that's right, you already have and even Dig is excited cause he claims that his dual-IPs are blocked here and he can't access NV News.
It's just too bad that trolls like you aren't banned here. You and the rest of them have ruined the B3D forums, hopefully you won't ruin any others.

OWA
03-04-06, 03:44 PM
crank up the resolution and anti aliasing and find out.
I can't go any higher than the 1920x1200 I was using but I can go higher with the AA. I'll try to redo things with 8xS or just use one card.

rwolf
03-04-06, 11:41 PM
Wow rwolf thanks for taking the time to come here and inform us. Now run back to B3D and arouse the ATI crew there with tales of your adventures here, lol. Oh that's right, you already have and even Dig is excited cause he claims that his dual-IPs are blocked here and he can't access NV News.
It's just too bad that trolls like you aren't banned here. You and the rest of them have ruined the B3D forums, hopefully you won't ruin any others.

Thanks. This is why I love posting here. I have Nvidia hardware in both my PCs by the way. :)

tertsi
03-05-06, 01:52 AM
I did rerun my tests with AA 4x... here are the results:

Doom 3
----------
16x12 + all maxed (no 3mood tweaks) + timedemo demo1

AF 16x + quality + all optimizations 56fps
AF 16x + quality - all optimizations 54fps
AF 16x + high quality + all optimizations 56fps
AF 16x + high quality - all optimizations 53fps

:: Overall: 3-5% performance hit.

Quake 4
----------
16x12 + all maxed (no Quak4 tweaks) + timedemo testi1 (singleplayer demo)

AF 16x + quality + all optimizations 60fps
AF 16x + quality - all optimizations 56fps
AF 16x + high quality + all optimizations 57fps
AF 16x + high quality - all optimizations 53fps

:: Overall: 5-11% performance hit.

16x12 + all maxed (no Quak4 tweaks) + timedemo testi2 (multiplayer demo)

AF 16x + quality + all optimizations 116fps
AF 16x + quality - all optimizations 109fps
AF 16x + high quality + all optimizations 110fps
AF 16x + high quality - all optimizations 103fps

:: Overall: 5-11% performance hit.

CoR
----------
12x10 + all maxed + timedemo testi1 (custom demo)

AF 16x + quality + all optimizations 60fps
AF 16x + quality - all optimizations 56fps
AF 16x + high quality + all optimizations 57fps
AF 16x + high quality - all optimizations 56fps

:: Overall: 5-7% performance hit.

walkndude
03-05-06, 09:35 AM
rwolf, it only takes a minute to check your prior posts and see what your all about... both here and at b3d. I honestly feel sorry for you, as well as those like you, over there. Get a life, troll elsewhere.

rwolf
03-05-06, 03:51 PM
rwolf, it only takes a minute to check your prior posts and see what your all about... both here and at b3d. I honestly feel sorry for you, as well as those like you, over there. Get a life, troll elsewhere.

Believe what you want. No skin off my back. I have been at ATI fan since R300 and since I had nothing but problems with Nvidia hardware.

RejZoR
03-05-06, 04:32 PM
Believe what you want. No skin off my back. I have been at ATI fan since R300 and since I had nothing but problems with Nvidia hardware.

Maybe because you don't know how to use it properly...
I liked R300 series, but i'm even more satisfied with 6600GT from NVIDIA.
No more ****ty bloated software, built in overclocking, excellent tray controls, very good image quality and performance.
I can't really ask for more.

NoWayDude
03-05-06, 04:42 PM
Believe what you want. No skin off my back. I have been at ATI fan since R300 and since I had nothing but problems with Nvidia hardware.

And after this statement, should we ever believe your opinion as partial?
Or are you just confirming your bias to the Red guys?
Because there is a diference between trying to be objective, and being dowright biased.

Your statement proves the second one does it not?

Vapor Trail
03-05-06, 05:24 PM
Hmmm, I run everything at the highest quality mode and can't be bothered with which card has 2-3 more FPS. Better things to do with my time frankly.

saturnotaku
03-05-06, 06:04 PM
Believe what you want. No skin off my back. I have been at ATI fan since R300 and since I had nothing but problems with Nvidia hardware.

Ugh, more dumb**** R3D trailer trash. Get the crap off this forum. :tool:

OWA
03-05-06, 06:28 PM
crank up the resolution and anti aliasing and find out.
I guess I still don't understand how to test it. Even after going to 8xS and a single card, I'm not seeing a difference.

Doom 3

61.2 vs 61.1 4xAA/8xAF through D3 (not forced in CP)

24.4 vs 24.4 8xSAA forced in CP, 8xAF set through D3

60.7 vs 60.7 4xAA forced in CP, 16xAF forced in CP

I've got global settings set to High Quality and in the game specific profile I've got high quality selected with each optimizations specifically checked and set to off vs each optimization left unchecked but still showing off. Is there something else I'm supposed to do like trying to force the optimizations on while I still have Hiqh Quality selected or something like that? Or maybe, have the global settings at Quality?

Edit: Having global set to Quality didn't change the one result I checked so I don't know. I give up unless someone can spot what I'm doing wrong. I'm getting the same results no matter what so who knows...

rwolf
03-06-06, 12:39 AM
Ugh, more dumb**** R3D trailer trash. Get the crap off this forum. :tool:

12 year old. I was posting here when you were in diapers.

rwolf
03-06-06, 12:42 AM
I can't go any higher than the 1920x1200 I was using but I can go higher with the AA. I'll try to redo things with 8xS or just use one card.


Say, how do you like your Dell 2405FPW 24" LCD. I just bought one.

rwolf
03-06-06, 12:45 AM
I did rerun my tests with AA 4x... here are the results:

Doom 3
----------
16x12 + all maxed (no 3mood tweaks) + timedemo demo1

AF 16x + quality + all optimizations 56fps
AF 16x + quality - all optimizations 54fps
AF 16x + high quality + all optimizations 56fps
AF 16x + high quality - all optimizations 53fps

:: Overall: 3-5% performance hit.

Quake 4
----------
16x12 + all maxed (no Quak4 tweaks) + timedemo testi1 (singleplayer demo)

AF 16x + quality + all optimizations 60fps
AF 16x + quality - all optimizations 56fps
AF 16x + high quality + all optimizations 57fps
AF 16x + high quality - all optimizations 53fps

:: Overall: 5-11% performance hit.

16x12 + all maxed (no Quak4 tweaks) + timedemo testi2 (multiplayer demo)

AF 16x + quality + all optimizations 116fps
AF 16x + quality - all optimizations 109fps
AF 16x + high quality + all optimizations 110fps
AF 16x + high quality - all optimizations 103fps

:: Overall: 5-11% performance hit.

CoR
----------
12x10 + all maxed + timedemo testi1 (custom demo)

AF 16x + quality + all optimizations 60fps
AF 16x + quality - all optimizations 56fps
AF 16x + high quality + all optimizations 57fps
AF 16x + high quality - all optimizations 56fps

:: Overall: 5-7% performance hit.


Here is the big question. Is the difference noticeable? Not the performance, but the quality.

tertsi
03-06-06, 02:07 AM
Here is the big question. Is the difference noticeable? Not the performance, but the quality.

56k warning....

http://www.skenegroup.net/tertsi/doom3/hqnoopt_vs_hqopt.html

rwolf
03-06-06, 02:11 AM
56k warning....

http://www.skenegroup.net/tertsi/doom3/hqnoopt_vs_hqopt.html

Well you can see it , but the real question is do you notice it while playing. You would think the jaggies would be more noticable when moving.

It looks great after the fix though.

tertsi
03-06-06, 03:23 AM
Well that depends on the game (e.g. is game fast action shoot'em up and game features + data) and in-game settings (e.g. gamma and brightness).

btw. I found that "High Quality" mode does deactivate all of optimizations in OpenGL except "Anisotropic mip filter optimization".

rwolf
03-06-06, 03:31 AM
Well that depends on the game (e.g. is game fast action shoot'em up and game features + data) and in-game settings (e.g. gamma and brightness).

btw. I found that "High Quality" mode does deactivate all of optimizations in OpenGL except "Anisotropic mip filter optimization".

Looks like just a bug. And a small one at that.

gstanford
03-06-06, 05:01 AM
12 year old. I was posting here when you were in diapers.

:bs: Errr... why don't I believe you? :retard: See attached image... :ban: