:: Back to news index ::

GPU PhysX in Dark Void: comparison video

with 4 comments

Dark Void is interesting title – at least from GPU PhysX perspective. It’s using brand new APEX Turbulence module for dynamic real time fluid smoke and particles simulations, which is looking much more realistic and natural then usual SPH-based smoke we’ve seen, for example, in Batman: Arkham Asylum.

Hovewer, more immersive gameplay came at great cost – Dark Void is most demanding GPU PhysX game currently.  Consider dedicated GTX260 GPU as minimum requirement for High PhysX settings.

In general, there are four PhysX modes in the game:

PhysX – Off: standart physics stuff, calculated on CPU – ragdolls, some dynamic rigid bodies, some particles. Default level for console versions.

PhysX – Low:  adds some dynamic particles – chunks and debris from bullet impacts, physically simulated sparks for Hypercoil and Magnetar weapons,  additional shards for destroyed enemies. We’ve seen such effects before, in almost every single GPU PhysX game, but this time there is really huge amount of particles, and they are not dissapearing for like half a minute or so.

Remark: The Low setting is the only setting that can be used to compare CPU PhysX effects versus GPU PhysX.

PhysX – Medium: additionally enables turbulence particle simulation for Disintegrator Gun. Each vaporazed enemy emits 30 000 particles, and there can be several simultaneous effects on-screen at once. As  ragdoll moves and rotates it imparts linear and angular forces into the fluid simulation, thus each disintegration is unique.

This mode is NOT available for CPU calculations, you just won’t be able to enable it without appropriate Nvidia GPU.

PhysX – High: includes all additional physics effects, and adds dynamic fluid smoke for Jetpack in Hover mode. Smoke simulation uses up to 100 000 particles, which are emitted every time whenever there is a change in applied power (i.e. the jetpack is turned on or off). This mode is NOT available for calculations on CPU.

Let’s talk a little about performance. First benchmarks are showing that APEX Turbulence simulations are very computational intensive. On our system (GTX 275 + GTX 260 PhysX) and High PhysX settings fps was jumping from 70 to 30 depending on scene complexity, sometimes – even lower.

Well, our work is done, now it’s up to you to decide, are those effects worth hardware they require. Later on be sure to visit Dark Void PhysX info mini-site, as we’ll add comparison PhysX screenshots soon.

Update: PhysX patch and benchmarks available. CPU utilization graphs included.

Written by Zogrim

January 18th, 2010 at 7:46 pm

Posted in PhysX Games

Tagged with ,

4 Responses to 'GPU PhysX in Dark Void: comparison video'

Subscribe to comments with RSS

  1. An additional GTX260 (or higher) für this gimmick?

    “are those effects worth hardware”
    For these some games out there? Definitely no!
    Only for those wo also buy SLi Configs for many many bucks more to have more AA/AF and Details than a Single Card can handle. And i think most of those don’t, too. For these few games?




    19 Jan 10 at 3:42 pm

  2. How about a performance comparison with different dedicated GPUs? I really hope to see such article soon.




    20 Jan 10 at 5:25 am

  3. GenL
    I have only two GPUs, so I won’t do any benchmarking soon, but I’ll try to gather other articles, same as I did with Batman.




    20 Jan 10 at 7:32 am

  4. one side of the screen looks like a bad game. the other side looks like a bad game with smoke, and better sparklies.

    has nvidia gone back to allowing those with ATI cards to run PhysX on their second card (likely their old Nvidia)? until they do, and possibly not even then, the A+ titles aren’t going to waste time doing extra coding for 30-70% of the market depending on which company has the best card at the moment.




    20 Jan 10 at 10:26 am

Leave a Reply

Refresh Image
Copyright © 2009-2014. | About project | Privacy Policy
PhysX is trademark of NVIDIA Corporation