There exist 2 modis, in the menubar. The first mode is the "ManageDB mode". It contain all CRUD-operations (create, read, update, delet) for solvers and instances. The second mode is the "Experiment mode". It manage the experiments.
Workflow:Start the EDACC-GUI.
Connect to the database and optionaly create the tables if they not exist.
Select the "Manage DB mode".
Stay in the "Solvers"-Tab. Press the "new"-button on the left (solver) side.
Declare a name for your solver and a description in the fields.
Press the "Add binary"-button and add the compiled binary of the solver to the database.
Optionaly press the "Add source files"-button and add the sourcecode and so on to the database.
If your solver has parameters, add them now on the right (parameter) side. Click on the "new"-button and declare a name (or short description), the prefix for the comandline and the order of the parameters on the comandline.
If all solvers with their parameters are added, press the "Save to DB"-button and switch to the "instances"-tab.
Press the "Add"-button on the right side to add new instances. Select the option "Automatic source class generation", set the depth of recursive folders and press the "OK"-button. Select the root-folder of your instances and EDACC will automatic add all instances in all recursive folders until the depth is reached.
Now everything to create an experiment were added to the database. Change the mode to the "Experiment mode"
Press the "Create"-button to create a new experiment. Declare here a name and a description for the experiment.
The new experiment should be selected and then press the "Load"-button.
Declare the "Experiment Parameters" on the bottom.
Switch to the "Solvers"-tab. Select the solvers to test and press the "Choose"-button. Declare the parameters you need and set a value.
Press the "Save"-button and switch to the "Instances"-tab.
First select one or more instaceclass oder userclass. Than select the instances to test the solvers and press the "Save"-button.
Switch to the "Generate Jobs"-tab and declare the number of runs per solver and instance. Select a queue and generate the "cluster package". At last "Generate Jobs" for the database.
Now everything exist. All solvers and instances were in the created package and the jobs exist in the database. Copy the package on your device or folder where the skripts can access to (on the grid).
Start the created skripts. In the next "Experiment mode"-tab (it is the "Job Browser"-tag) the progress is shown.