In addition to the points mentioned in the General Performance section, the following should be taken into consideration when rendering on the CPU.
CPU rendering is implemented in a multi-threaded manner. The number of threads used is configurable in the options dialog or on the command line with the —nthreads option. Using all cores on your machine will allow the render to finish earlier, at the expense of UI responsiveness. If you’d like a more responsive UI, you have two choices:
Use a number of threads that is less than the number of cores available.
Lower the priority of the threads via the options dialog, or the —priority option on the command line.
The calculations for rendering can be done using 32-bit single precision, or 64-bit double precision numbers. These are configurable on the toolbar with the SP/DP buttons, or on the command line with the —bits=33, —bits=64 options. Unless you have an extreme need for speed, or have a low power machine, you should always use double precision.
Most flames look the same with either setting. However there are some where precision matters greatly and will thus produce completely different outputs between the settings. You can save yourself time switching back and forth by just always using double precision on the CPU.