Towards one million flocking boids using CUDA (real-time)

Loading...

Sign in or sign up now!
Alert icon
Upgrade to the latest Flash Player for improved playback performance. Upgrade now or more info.
3,751
Loading...
Alert icon
Sign in or sign up now!
Alert icon

Uploaded by on Nov 23, 2008

Results (update stage only):
- 2^16 (65K): 47 fps;
- 2^17 (131K): 29 fps;
- 2^18 (262K): 14 fps;
- 2^19 (524K): 6 fps.

Simulation details:
- Full-GPU, nVidia CUDA 2.0;
- Each boid looks at 7 nearest neighbors within a search radius equal to 9;
- OpenSteer compatible;
- Guaranteed high precision behavior computation, comparated with the original Reynold's implementation on OpenSteer.

Hardware details:
- Intel Core 2 2180;
- 2Gb RAM;
- nVidia GeForce 8800GTX 768Mb.

ISISLab - University of Salerno - Italy

Category:

Science & Technology

Tags:

License:

Standard YouTube License

Link to this comment:

Share to:

Uploader Comments (ber85it)

  • GTX 295 SLi needed?

  • 1 X 8800GTX

see all

All Comments (7)

Sign In or Sign Up now to post a comment!
  • Damn, that's 10^3 times more boids than i can run on a quad-core while maintaining over 3 fps average. Got some pretty big overhead for collision detection, bounds and predators though.

  • fuckin generic musc man

  • or gtx285 3 way sli!!!!!

  • Awesome stuff! Great music btw - which song is played in bg?

  • image what you could have done with a 295SLi

  • Impressive! 512K boids and a decent 6 FPS (regarding the HUGE number of boids)!!!

    I sent you a message did you read it?

    Regards,

    Nad.

Loading...
Alert icon
0 / 00Unsaved Playlist Return to active list
    1. Your queue is empty. Add videos to your queue using this button:
      or sign in to load a different list.
    Loading...Loading...Saving...
    • Clear all videos from this list
    • Learn more