OpenCL Flocking Demo

A demo showing the performance difference between the CPU and GPU in calculating the movement of a flock of boids and predator. The program is multithreaded for up to 32 threads when using the CPU. OpenCL is used when in GPU mode enable processing on the video card.

This demo was run on an AMD 1090T and ATI HD 5850. Also tested on my i7 laptop with a Geforce 425M card and it ran fine at about half the performance as my desktop.

The upper left shows the frames per second, if it is was using the CPU or GPU and how many nodes or particles are in the flock. The number of threads is also displayed and changeable when operating under the CPU mode.

The demo was programmed in C#. SlimDX was used for the graphics interface. OpenCLTemplate and Cloo was used to enable OpenCL acceleration.

Category:

Science & Technology

Tags:

• OpenCL
• C#
• SlimDX
• Flocking
• Swarm
• AI
• GPGPU

[AzteKpr1de]

OpenCL is the future.

[webDes33]

@AzteKpr1de really I hope it will be use for games!!

[rich1051414]

@DarkGhostlyDeath while it's true the processing is more dedicated, this isnt what causes the performance difference. It has to do with the number of calculations it can do simultaneously. while a 6 core can do 6 at the same time, gpu's can do many many more than that at the same time.

[DarkGhostlyDeath]

GPUs are pretty dedicated for the task they perform however CPUs are pretty generalized even with task specific instructions.

[regemo]

I guess a clear example of the parallel processing power of gpus for simple calculations compared to the more serialized cpu processing.