To start a parameter study in the cloud, create a GeoPython Macro containing variables. Then go to Macro→Execute Macro / Script and navigate to your macro file. Enable Vary and go to the Parameters to define the variation for each variable. Then click on Run in Cloud.
If you are not logged in, the Login to GeoDict Cloud Service window opens. Enter your Username, Password, and your One-time code from the authenticator. Click OK to continue to the Run GeoPython Macro in GeoDict Cloud dialog. If you are already logged in, this dialog opens directly.
Run GeoPython Macro in GeoDict Cloud Dialog
The Run GeoPython Macro in GeoDict Cloud dialog offers several settings:
The Project Name allows you to group multiple simulations together by giving them a shared project name. The default name originates from the name of the currently selected project folder, e.g., MyFirstGeoDictProject.
The Job Name Prefix serves to identify the specific simulation that is run. The default name consists of Macro execution and the name of the macro. The name of the individual jobs in the parameter study will be extended by the corresponding variation of the variables.
When this option is activated the structure that is currently loaded in GeoDict is uploaded to the cloud machine together with the job. This is required for commands which operate on the current voxel structure. This applies to most Analyze and Predict modules. Therefore, this option is active by default when a structure is currently loaded in GeoDict.
When this option is activated the entire current project folder is uploaded to the cloud machine together with the job. This can be necessary if additional files in the project folder are required to run the computation, e.g., for continuing a FlowDict simulation from a previous result. We suggest preparing a new project folder for this purpose, containing only the required data to avoid uploading unnecessary files.
Select the GeoDict Version that you want to use. You can only select the same major version as you are running locally, e.g., GeoDict 2026. However, you can choose between different service packs, if they are available on your cloud environment. Note that this allows you to, e.g., test a new Service Pack release before installing it locally.
Here you can select which type of cloud machine you want to use for the simulation. You can see the Number of (#) cores, the RAM, and what type of GPU is available. To see cloud machines with one or more GPUs, enter a value in the Number of GPUs box. GPUs are necessary to effectively use AI functionalities in GeoDict (e.g., GeoDict-AI, FiberFind-AI). Click in the checkbox of the machine type that you want to use for this job and then click OK.
Note! Be aware that the Number of Cores shown includes cores available through the use of hyperthreading. That means, a machine with 32 logical cores consists of 16 physical cores. Some GeoDict functionalities cannot efficiently use hyperthreading. Their speed is limited by the ability of the processor to process large amount of data and not by the ability to do many arithmetic operations. Therefore, if you select a machine with 32 cores from the list, the runtimes may be similar for using 16x parallelization or 32x parallelization, as cores available through hyperthreading may not add a significant speedup.
This option allows you to define a timeout for the job in hours. If the job has not completed after the given time, it is automatically terminated. The maximum value is 720 h (30 days). If you do not want any timeout, set this value to 0. Then the simulation runs until it is completed, terminated, or crashes. Setting a timeout can be useful to avoid additional costs, e.g., for simulations that fail to converge.
Here you can see all variations of the variables in your macro shown in a table. Each row corresponds to an individual job that will be executed by a single virtual machine. If your cloud setup has a limit on the number of simultaneous instances, make sure that the parameter study does not create more jobs than currently available instances. In the shown example, the number of jobs created for the parameter study is 4, while the limit of simultaneous instances is 5. If you submit more jobs than that limit, you will receive a error message during the upload. Surplus jobs will be shown as Ready in the GeoDict Cloud Projects dialog, but will not start once another job is completed. Delete these jobs and wait for another jobs to finish before resubmitting the simulation.
After defining all settings, click Run to send the jobs of the parameter study to the cloud. You will then see a dialog showing the progress of packing the jobs and sending them to the cloud. All your custom GeoDict settings (e.g., your material database) are also automatically packed and sent together with the job. This ensures that the same settings are applied for local and cloud machine simulations. Note that all submitted jobs are executed at the same time in parallel on multiple machines.
After the upload of all jobs is completed, the GeoDict Cloud Projects dialog opens automatically. Here you can see the current status of your submitted jobs as well as all other jobs that you have in the GeoDict Cloud.
Press Refresh (or F5) to update the view. You can close the dialog and continue to work with GeoDict locally. Open the dialog any time by selecting Cloud & Queue→GeoDict Cloud Projects in the menu bar. When a simulation of the parameter study is completed, the virtual machine shuts down automatically and the status of the job changes to Completed. You can now download the results.
