Screenshot 2

Changed the rendering of particles a bit. The screenshot shows the grid boundries and particles initialized.

update 04/26/2010

Update since last post:

- Particle neighbor array construction
- Per particle pressure and density calculations
- Using the pressure and density values per particle, calculation of forces

Next Steps:
- Velocity and position update steps (hopefully on GPU!)
- Lots of debugging

Problems faced till now:
-Debugging on the GPU is a pain without the ability to look at the values on the card!
- Initializing a struct variable on the card from the host doesn't seem to work for me. So for now using hard coded values for the parameters in the cu file.

Screenshot 1

Visualization of the particle grid when initialized.
Next Steps:
Compute neighbors for each particle to be used in the simulation.

Project Update 4/20/2010

Apologies for the late update!

I have setup the static particle grid for SPH simulation till now. The grid shows a color gradient across the grid with respect to the distance of particles from the boundaries. The rendering is setup so that I can visualize the output during the following stages of the development process. I have started coding the kernels needed for simulation to work on GPU. Currently I am working on the neighbor search algorithm on GPU as described in the SPH on GPU paper by Harada et al. This algorithm is the main building block for the simulation since a robust implementation of this would ensure an efficient simulation.

More updates to follow in the coming days!