Dark Void Benchmark and PhysX impact in performance

[evolucion8]

I downloaded and installed the Dark Void Benchmark, ran it and I set the resolution at 1280x0124 with everything maxed, the game doesn't have an Anti Aliasing option and the only way to use Anti Aliasing in the game is renaming the executable to UT.exe and forcing Anti Aliasing through the Catalyst Control Center.

Then I went to PhysX option and it states that I need a dedicated GPU for PhysX for medium settings or higher, I can only choose low, I ran it and I scored min 33fps and max 66fps, I found that the most demanding area dropped to 30fps. Then I deleted the PhysXCore.dll and the PhysXLoader.dll in the game's directory, suddenly my fps got a nice boost. It increased from 33fps to 44fps in minimum frame rates and from 66fps to 72fps in maximum frame rates, and the most demanding area of the benchmark never dipped below 42fps comparing to the 30fps that I got before.

Seems that deleting the PhysXCore.dll and PhyXLoader.dll increased the performance of my AGEIA card with Dark Void PhysX. Disabling the AGEIA PhysX acceleration using the Control Panel made the game unplayable when PhysX is On in the game. These are the results...

PhysX On:

Hardware PhysX before deleting files No AA/AF - Min: 33fps Max: 66fps
Hardware PhysX after deleting files No AA/AF - Min: 44fps Max: 71
Hardware PhysX after deleting files 8XAA/16XAF - Min: 41fps Max: 66fps

Seems that the results shows that I'm limited by the AGEIA card or by the CPU, but to me, the AGEIA card is the limiting factor, it's a PCI card.

Now with the AGEIA card disabled:

Software PhysX No AA/AF - Min: 4fps Max:16fps
Software PhyX 8XAA/16XAF - Min: 4fps Max: 14fps

Made no difference because I'm severely bottlenecked due to the lack of PhysX acceleration.

Now with PhysX disabled in the game:

NoAA/AF: Min: 166 Max: 200fps

8XAA/16XAF - Min: 91fps Max: 113fps

That's a hefty drop in performance, but considering that the game uses the Unreal 3 engine (Looks quite pathettic considering that the Unreal 3 engine is capable of excellent graphics), plus I had to rename the executable to UT3, force the Anti Aliasing through CCC which uses brute force Anti Aliasing and the fact that the game is relatively new, it bring space to driver optimizations later on. The game is relatively easy with today's cards and even with two generation behind cards like the 8800 series or the HD 2900/HD3800 series. Post your results and opinions!!

 

[Lonyo]

Does PhysX actually make any kind of meaningful in-game difference?

 

[toyota]

Does PhysX actually make any kind of meaningful in-game difference?

no. http://www.youtube.com/watch?v=BGCZtXg5LyA&feature=player_embedded

even on low the framerate will tank to well below 30fps at times on my gtx260. without physx it usually never dips below 60fps. yet another game that I would rather have better framerate in then to take the hit for such meaningless physx effects.

 

[Qbah]

no. http://www.youtube.com/watch?v=BGCZtXg5LyA&feature=player_embedded

even on low the framerate will tank to well below 30fps at times on my gtx260. without physx it usually never dips below 60fps. yet another game that I would rather have better framerate in then to take the hit for such meaningless physx effects.

The effects look kinda cool imo. If you have an nVidia card, you might as well turn it on. No need to have a game running at 100+FPS...

Now, getting a G200b card for this over the new HD5850/HD5870/HD5970's that will smoke any nVidia card in other, arguably better games? Hell no.

PhysX seems to lack cool games in general (and not just because they have PhysX). The only one is the Batman game. All other games range from medicore to completely crap. Dark Void, based on the reviews, is medicore at best - many calling it boring and crap in general. A 58% from reviews/3.9 from users score speaks enough about it for me.