Skip to content

Latest commit

 

History

History
75 lines (53 loc) · 1.61 KB

README.md

File metadata and controls

75 lines (53 loc) · 1.61 KB

An Analysis and Evaluation of Database Cracking Kernels

Licence

The source code is provided under the Apache v2 licence. A copy of the licence is provided in the LICENCE.TXT file, which is included in this repository.

Instructions

Download the Source Code

git clone https://github.com/BigDataAnalyticsGroup/RewiredCracking.git
cd RewiredCracking
git submodule update --init --recursive

or

wget https://github.com/BigDataAnalyticsGroup/RewiredCracking/archive/master.zip
unzip master.zip
rm master.zip
cd RewiredCracking
wget https://github.com/opcm/pcm/archive/master.zip
rmdir processorcountermonitor
unzip master.zip
mv pcm-master processorcountermonitor

Build the Project

mkdir build_debug
cd build_debug
cmake -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_BUILD_TYPE=Release ..
make -j 42

Mount the hugetlbfs

sudo ./mount_hugetlbfs.sh

Mount the cpuset pseudo filesystem

The following commands require privileged access.

mkdir /dev/cpuset
mount -t cpuset cpuset /dev/cpuset

Execute the Benchmarks

Use build_debug/bin/partition -h for help.

build_debug/bin/partition 4 DENSE

You find the results in a CSV file that is named after the name of the host, e.g. partitioning_MY_AWESOME_HOSTNAME.csv.

Generate the Plots

The plot script is written in R. Make sure to have all required libraries installed.

R -q --no-save --file=partitioning.r --args partitioning_MY_AWESOME_HOSTNAME.csv

Explore the Results

The plots are named partitioning_MY_AWESOME_HOSTNAME_<NAME>.pdf.