This tutorial was generated from the file projects/Harvey2015/test/TestMemoryUseLiteratePaper.hpp at revision r23521.
Note that the code is given in full at the bottom of the page.
This class was used to produce the results in Figure 4.
It constructs a large cell population and measures the amount of memory
in use. By running on larger numbers of processes, a proportionally
smaller amount of memory is used by each process.
The geometry for the construction of the population is contained in
This test suite should be run in parallel. It should be run on several numbers of processes (from 1 to 32 processes in figure
in the paper.)
A useful for loop (in bash) would be
This runs the test on increasingly larger number of processes and only outputs the lines which state
the individual memory use. Some post-processing will be needed in order to select the maximum memory use to
plot on the vertical axis.
This function prints the memory usage at the time it is called.
rPrefix contains the rank (identifier) of the process
This class was used to produce the results in Figure 4.
It constructs a large cell population and measures the amount of memory
in use. By running on larger numbers of processes, a proportionally
smaller amount of memory is used by each process.
Record the rank of each process so that it can be output
Make a NodesOnlyMesh in which to store the nodes which are to be read from file
There is a 1024 y-range over which this mesh is split. This means that each of 32 processes will take a y-range of 32.
The maximum interaction distance will define the size of the indexing boxes and thus define
the size of the strips which are the unit of parallel distribution
Here we call a parallel helper method which reads the cell locations
from a file on disk. This is a uniformly spread rectangle which is 1000 cells in the
x direction and 1024 cells wide in the y direction.
The proliferative type ensures that the cells are not growing, although this information is never
used. No simulations are run.