This project calculates the mandelbrot set with arbitrary percision floating point numbers and distributes the work using MPI.

Please see the report for output samples.

Dalton Caron dpcaron@csu.fullerton.edu Jeevika Yarlagadda jeevikayarlagadda@csu.fullerton.edu

The user must have GCC, libgmp, and make installed to compile the program. The user's compiler
must comply to POSIX standards. To compile the program, use the Makefile as shown below.

make build

This command produces the a.out executable file.

This program produces bitmap image files and binary files storing the parameters used to produce the images. There are a variety of command line options:

• -e: save the mandelbrot bmp every zoom iteration of the algorithm
• -f [file]: load parameters from a binary save file
• -z [number]: the amount of times the algorithm will recompute the set with a deeper zoom
• -s [number]: set the size of the output image. Must be a power of 2.
• -i [number]: the amount of times to iterate over the formula f(z). Must be larger the deeper the zoom.

An example command is shown below.

mpirun --mca opal_warn_on_missing_libcuda 0 --oversubscribe -n 5 a.out -i 1000 -z 3 -s 256  

Depending on the parameters, the algorithm may execute in a few seconds or few hours. The default parameters in the Makefile run in a few seconds on modern machines. The Makefile default is executed using the following command.

make run

There are debugging presets, but please disregard them unless you are a developer.

The pseudocode for this project is located in report/report.pdf in the root of this repository.