Skip to content

Comparing the Dune-ISTL and Ginkgo libraries against each other

License

Notifications You must be signed in to change notification settings

Ben90001/BA-Ginkgo_vs_Dune

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

38 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

BA: Ginkgo vs Dune

Comparing the Dune-ISTL and Ginkgo libraries against each other

Versions

Using DUNE 2.9 and Ginkgo 1.8.0

Installation

Dune

To install Dune first adjust parameters within dune_installer.sh afterwards install as follows:

./dependencies/dune_installer.sh DUNE
cd DUNE/
./buildmodules.sh

Afterwards you might want to remove the modules common and istl from the buildmodules.sh build process, so that upon future calls only the added project-module gets recompiled.
Add your Code to dependencies/DUNE/dune-evaluation/src
Find your binary at dependencies/DUNE/release-build/dune-evaluation/src

Ginkgo

To install Ginkgo simply run:

./dependencies/ginkgo_installer.sh

NOTE: This fully builds Ginkgo, compilation is set to use 64 cores. The Spacke environment activation probably needs deactivation on other machines.

The script builds into dependencies/GINKGO/include.

Run Ginkgo

To run the Ginkgo experiments install Ginkgo as described above.
Afterwards compile the experiment as follows:

cd src/ginkgo
./rebuild.sh
cd ..

To run with only the needed input modify src/generate-results.sh accordingly. Simply comment out the ISTL runs within the execution loop if those aren't needed.
NOTE: The cpu/gpu flag indicates not whether cpu or gpu is used but whethere matrix_data or matrix_assembly_data is used. Results are outputed into src/results/.

About

Comparing the Dune-ISTL and Ginkgo libraries against each other

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published