![]() When GPU memory is exceeded, TurbulenceFD switches back to the CPU on-the-fly. All features are supported at the same quality. Unlike some GPU based tools, this is not just a stripped down version of the CPU simulation. It features a hybrid CPU/GPU simulation pipeline that achieves enormous speedups. And there’s a simple reason for that: today’s high-end GPUs have 8-15 times the memory throughput of high-end CPUs. ![]() Yes, twelve times! 10 minutes instead of 2 hours. To the artist this means that more iterations can be run in less time, making the work with fluids more intuitive and productive. And implementing this pipeline using the latest High Performance Computing technology to optimally exploit Memory Caches, Multi-Core CPUs and advanced vector instruction sets. ![]() This includes a careful selection of efficient numerical methods that provide high accuracy and stability throughout the simulation pipeline. That is why TurbulenceFD’s simulation pipeline has been designed from the ground up to optimize performance. The biggest technological challenge in fluid simulation is the handling of the large amounts of data that a sequence of voxel grids requires. The flow then carries along these emissions in a physically plausbile way that creates the realistic look of fire, explosions, vapor, clouds, dust and much more. To setup a fluid simulation, the artist uses any type of geometric object or particle system to paint the sources of smoke, heat, fuel, etc.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |