Precalculating the Mandelbrot set - because it's there
This is mostly a learning exercise for me.
- Key things:
- use of anonymous unions and NAN for efficient storage of the mandel_X.dat files (depite the fact that they are rather large still...) This was my original 'cool idea' / excuse for starting this.
- use of mmap in a 64 bit address space for basically doing anything. Nice.
- use of pthreads to utilise multi-core processors
- And of course, the key thing:
- using a computer for it's intended purpose: computing, taking all the resources it possibly can in the process :-)
- For the future:
- distributed parallelism
- not only incremental in terms of iterations, but also incremental in spatial resolution (which will probably require a change in the file format, or something like a group of 4 mandel_X.dats which are postprocessed to create a mandel_(X+1).dat, by storing an offset of zero or 2^-(X+1) in each of the real and imaginary axes...
- an actual command line interface, with options and everything