Release 0.3.5: first introduction of OpenCL and GPU computation.

In this release I’ve introduced OpenCL and GPU in the development of PyParticles, and the firsts result are very well , try by yourself the galaxy demo.

The acceleration respect the the simulation based only on numpy is impressive.

Galaxy

If you want you can try also the fountain demo were I’ve introduced OpenCL in the drag computation, and the result is very encouraging and now you can simulate easily up to 700K particles or 1M particles.

As second point I’ve fixed the VSYNC problem, but only under windows.

New version 0.3.4 and VSYNC

For this version I’ve rewritten the logging system, now id more flexible and fast, test it with the solar system demo, I’ve also added the FPS counter and I’ve also improved the documentation.

 

The galaxy demo above will be available with the next release (0.3.5) stay on-line:

For the next version I’ve started to works with OpneCL and PyOpenCL

Disable VSYNC

Some video card has enabled the vsync, at the moment this option slow down the simulation in PyParticles if you use OpenGl, so if you want the maximal velocity you must disable this option:

Under Linux

with NVidia binary driver you must start the Nvidia server settings [command: nvidia-settings], select OpenGL Settings and uncheck Sync to VBlank

  • For a generic user and AMD card, try to watch this video:

Under Windows for NVidia and ATI users, follow the instruction at the link bellow:

http://windows7themes.net/how-to-disable-vsync-in-windows-7.html