Requires Python3, and the following dependencies, installed via pip
:
pip3 install matplotlib pandas seaborn
Run make
to produce all the figures (produced as PDFs).
The plotting scripts read from all the .csv
files in this directory,
that need to manually be moved in here from the benchmark
outputs:
./data.csv
and./prep_time.csv
contain outputs from../benchmarks/
../microbench-mem-access
contains output from../microbenchmark-compare-read-arr/
. You will likely need to tweak themachines
map in./microbench-mem-access.py
for it to pick up the data from your machine. Currently it is set to the machine names that were used for evaluation in the paper.
The raw data files used to produce the graphs in the paper are already included in this directory.