Hardware Advice for Faster Simulating
Here's some tips to help you make sure your Phoenix sims calculate as fast as possible.
[TABLE="align: left"]
[TR]
[TD]Hardware[/TD]
[TD]Recommendation[/TD]
[/TR]
[TR]
[TD]Bus speed[/TD]
[TD]Phoenix transfers huge amounts of data between memory and the CPU. The entire simulation grid gets processed tens to hundred times per simulation step, so the bus speed is the bottleneck in most simulations.[/TD]
[/TR]
[TR]
[TD]CPU speed[/TD]
[TD]Processor speed is more important than core count. Also, NUMA and multiprocessor architectures do not perform well with Phoenix. Simulating on one NUMA node is often faster than waiting for multiple nodes to synchronize.[/TD]
[/TR]
[TR]
[TD]Fast Storage[/TD]
[TD]Both large and small simulations take significant time just to write the cache files from the simulation and to read them back for the preview, so good storage speed can help decrease simulation times up to 30%. Turning off the preview for all grid and particle channels during large simulations could also help speed up the simulation.[/TD]
[/TR]
[TR]
[TD]GPU[/TD]
[TD]GPUs are currently only used for the GPU preview of fire/smoke simulations.[/TD]
[/TR]
[/TABLE]