Things have been a little too busy lately, but I managed to get my NSE Water Simulation updated to allow a grid based processing scheme. This allows the grid size to be significantly larger while still utilizing future GPU's hardware during processing. An updated video of the simulation can be seen here, which uses a 128 x 128 grid as opposed to the 32 x 32 grid from my previous video.
Here is also a still image of the simulation as well:
The processing elements are 16 x 16 groups, with a total of 8 x 8 groups in the simulation. The colors of the grids are alternated so that you can see what parts of the simulation are calculated together and which ones aren't. Overall, I am quite pleased with the final method, which also incorporated changes to the memory access patterns to reduce any thread contention.
I'm currently working on a full article discussing the implementation of the method and how it works, although it will likely be a few days or a week until it is finished. I'm currently unsure of how to get the article out there at the moment - I may publish it here, submit it to GDNet as an article, or use a mysterious option #3. I should decide in the next few days what avenue I will use, and I'll post here about what I come up with.
My plans for June are primarily to wrap up my thesis (I can't believe I'm nearing the end!) so my other development projects may be sparsely updated. However, after the thesis is done, I plan to make a strong push on my D3D11 renderer and to develop a few more samples for it so stay tuned!