Unreal Tournament 3 - Nvidia CP-Settings Performance & Quality Review

[Jelle Mees]

System Information

Operating system: Microsoft Windows XP SP3
System memory: 4 GB 1Ghz
CPU: Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
CPU speed: 3600 MHz
Sound system: Aurilium Sound Agent 2

VGA Information

Graphics card: NVIDIA GeForce 8800 GT @ 756Mhz / 1890Mhz / 2000Mhz ( 1.1V VoltMod )
Memory: 512.0 MB
Driver version: 6.14.0011.7739 (English)
PhysX Support: Enabled ( 08.06.12 )

Game Configuration:
2048x1536 - Highest Settings

--------------------------------------------------------------------------------------

Anisotropic Filtering

The performance impact when you turn on Anisotropic Filtering:
- 2x: 1%
- 4x: 2%
- 8x: 6%
- 16x: 7%

If you want to use AF, just use 4x. The difference between 4x,8x and 16x is hardly noticeable while playing. The difference between OFF, 2x and 4x is pretty big. If you use 4x you will only have 2% performance impact and everything will look more detailed.

Antialasing Mode - Gamma Correction

No quality or performance impact.

Antialiasing setting - Mode

Some review websites zoom in on screenshots to spot the difference in image quality. That's very stupid because nobody plays zoomed in ( unless you are sniping offcourse ).

Performance impact when you turn on Antialiasing:
- 2x: 17%
- 4x: 37%
- 8xQ: 57%
- 8x: 41%
- 16x: 42%
- 16xQ: 62%

Whatever you do, avoid the Q settings. The performance impact is just to high. I don't really see the point of using AA in this game. There is so much activity in this shooter, you won't notice if an edge isn't 100% smooth here and there. Just turn up the resolution and graphics as much as you want. With a 512MB card you can go up to 2048x1536. Only turn on AA if you are using an ATI 3870X2.

Antialiasing - Transparency

No quality or performance impact.

Conformant Texture Clamp:

No performance impact. If you turn this setting on then it will increase the light that comes from light-sources.

Force Mipmaps:

No quality or performance impact. The reason why you see the lights different on every screenshot is because the lights change constantly.

Texture filtering - Anisotropic Sample Optimization:

Performance improvement with Anisotropic Sample Optimization: 4%

If you turn on Anisotropic Sample Optimization, your framerate will be 4% higher. In other words. You can turn on 4x Anisotropic filtering and if you enable this setting, your framerate will be 2% higher! You don't have to doubt about this one, just enable this setting! Extra performance without quality loss, why not?

On the right there is some smoke in one of the screenshots, ignore that. The smoke comes out of the vents every now and then.

Texture filtering - Negative LOD Bias:

No difference in quality or performance. Let's move on to the next setting...

Texture Filtering - Quality:

Only "High Quality" has a negative performance impact but as you can see, there is no better quality. Just put this setting on "High Performance".

Texture Quality - Trilinear Optimization:

According to Nvidia, this setting on "On" should give better performance and reduced quality. But in reality there is no quality difference at all and the performance is 4% lower! I tested this on 3 different maps to be 100% sure because I didn't beleave it at first. Every UT3 and GOW gamer has to disable this setting.


Threaded Optimization:

This setting has no performance impact. Just leave it at "Auto".

Vertical Sync & Triple Buffering:

These settings have no impact on the graphics at all. Normally they would have a performance impact but that's not the case in UT3. Some gamers play with Vsync and enable Tripple buffering to compensate the performance impact, others just play without Vsync for the best performance. Well, in this game, everyone can play with Vsync enabled WITHOUT the need for Tripple buffering. Vsync has no performance impact at all in this game.

--------------------------------------------------------------------------------------

Since June 2008 it's possible to enable hardware PhysX on all G80,G92, GT200 and future Nvidia products. I benchmarked on 15 UT3 maps on 1600x1200 with the default Nvidia "High Quality" CP-settings with Vsync disabled. I calculated the avarage framerates:

PhysX Enabled: 116FPS
PhysX Disabled: 92FPS

Performance difference: 21%

--------------------------------------------------------------------------------------

Conclusion:

With a Nvidia G80/G92/GT200 Nvidia product and the correct settings you can have 29% extra performance without any quality loss! Now, 4xAA has 37% performance impact. Isn't it nice that with a Nvidia G80/G92/GT200 Nvidia product and the correct settings it's only 8%? I sure think it is! Remember at all times that this Performance/Quality-Review is only good for the games "Unreal Tournament 3" and "Gears of War".

I hope you all liked this performance/quality review. Expect many more in the near future.

[EviLZeD]

great review im going to try tweaking my ut3 now

[WhiteLotus]

this is a very good idea - shows what the the AA and such look like, so can actually compare easily - nice job.

[das müffin mann]

ya great review, ill tweak it when i get home

[Dangle]

How do you turn on Anti-Aliasing? Is this only possible through external software?

[ktr]

Sadly, UT3 is not getting much attention. Atm, there are only like 600 players. UT2k4 still has 4500.

[Jelle Mees]

Originally Posted by Dangle

How do you turn on Anti-Aliasing? Is this only possible through external software?

You have to use your driver settings. But this review doesn't apply to ATI hardware. The performance and quality impact is totally different.

Originally Posted by ktr

Sadly, UT3 is not getting much attention. Atm, there are only like 600 players. UT2k4 still has 4500.

Epic wasn't smart enough to release some HD trailers a week before the release. UT3 is an awesome game but it didn't get advertised enough. It was such a hype years before the release. I guess people just forgot about the game.

Friend of mine lauched an UT community for Belgian UT gamers, we had to shut it down already just because it was to hard to find players.

[Jelle Mees]

Updated this review due to the recent PhysX support on Nvidia hardware.

[wolf2009]

nice , thx .

[Jelle Mees]

Lol, I posted "35", I am not crazy. I only benchmarked 15 maps. Edited post

[xubidoo]

i thought PhysX support in the 8800/9800 series wasnt enabled until the drivers that come with the 9800gtx+ are released? which is in July.

[Jelle Mees]

Originally Posted by xubidoo

i thought PhysX support in the 8800/9800 series wasnt enabled until the drivers that come with the 9800gtx+ are released? which is in July.

I downloaded 177.39, then I downloaded a modded inf for 177.35, I edited the version line so that the inf works for 177.39, then I installed 177.39 with the modded inf and it worls fine

If you want I'll upload the modded driver for ya

[xubidoo]

thanks ,didnt realise beta drivers were out already that enabled it

wat do i need to DL to test it out ? im on xp32 btw.

[Jelle Mees]

Originally Posted by xubidoo

thanks ,didnt realise beta drivers were out already that enabled it

wat do i need to DL to test it out ? im on xp32 btw.

Download and install this ( modded 177.39 driver for XP x86 ):
http://rapidshare.com/files/124025145/177.39.rar

And download and install this:
http://nekrosoft.net/NVIDIA_physX_8_06_12_driver.exe

Let me know if it works for ya

[xubidoo]

seems the nekrosoft link is borked

looked around but that seems tio be the only place that had it

i cant even get the file from rapidshare ,ive always had probs with thiose stupid code things but to ask for letters witha cat... they all look the same to me heh ,i tried about 5 times and gave up.

thanks anyway jelle mees ,ill just have to w8 until its released properly.

[Jelle Mees]

Originally Posted by xubidoo

seems the nekrosoft link is borked

looked around but that seems tio be the only place that had it

i cant even get the file from rapidshare ,ive always had probs with thiose stupid code things but to ask for letters witha cat... they all look the same to me heh ,i tried about 5 times and gave up.

thanks anyway jelle mees ,ill just have to w8 until its released properly.

The nekorsoft link seems to work just fine here. And I just send this mail to rapidshare.com:

Today I received the 5th complaint about www.rapidshare.com . Apperently people without premium account need to tell how many cats there are in a CAPTCHA image or something and nobody is able to tell wich is a cat and wich isnt.
Please fix this as soon as possible.

Here you can find many more people complain about the new "cat-system":
- http://forums.superiorpics.com/ubbth...86/an/0/page/0
- http://forums.techpowerup.com/showpo...4&postcount=15
- http://www.labnol.org/internet/favor...a-images/3064/
- http://forums.bf2s.com/viewtopic.php?pid=2037786
- http://www.xboard.us/showthread.php?t=274283
- http://thewobblingmind.wordpress.com...pidshare-cats/
- http://www.fiveeightforums.com/rapid...x.html?t=54440

I can make the list a lot longer, but I am not gonna waiste my time doing it. You get my point.

My guess is you guys want to force people in to using rapidshare accounts. Well, let me tell you this. If you don't change the CAPTCHA before my account expires I will not extend my account AND I will no longer help people with my Paypal to buy Rapidshare accounts. There are plenty of hosts who are just as popular as rapidshare.com wich don't have this ridiculous CAPTCHA images.

If you did it to improve security, you have failed:
http://www.raymond.cc/blog/archives/...-by-cryptload/

Since I have Rapidshare premium myself and I help about 15 others to pay for it with my paypal, that's about 1400$/year they will be missing from me.

I know that my email wont do much, but maybe if they receive enough of them, it will change before my account expires.