With the amount of data that is now available to all geoscience professionals, the stress we are placing on our software and computing resources is increasing. The complexity of the projects being built in Leapfrog is reaching new levels with the increasing functionality being offered across the software. This complexity is placing the Leapfrog and user computing resources under significant stress. This is also creating ever larger files and projects that need to be stored on our local and cloud environments.
Seequent is committed to continually improving the processing and storage of Leapfrog projects and their associated data.
With the Central workflows now available to Seequent customers, the overall size of Leapfrog projects is more important than ever. Smaller projects mean faster upload times of projects to Central and just as important, less consumption of local disk space.
We have made changes to the compression and storage of several data types, that have resulted to big reductions in size on disk. The biggest gains are found in handling of image data, storage of octree block models and associated data, and the storage of array results that are produced from within Leapfrog. Projects that will benefit most from these changes are ones containing a large amount of imagery and those with large block models, but healthy gains should be seen across the board.
“The results of analysis conducted on five vastly different projects yielded reductions in overall project size ranging from 30 to 65% improvement.”
In one extreme case, a project that was over 210 Gb in size was reduced to ~ 30 Gb after upgrading to LF 2021.2 – an 86% size reduction!
We have improved existing algorithms by modernising the way we store, access and process information relating to meshes and volumes within Leapfrog.
One area where a significant gain has been made this release is to the process of selecting points falling inside mesh volumes. Users should see a significant speed improvement when filtering points used inside a geological or numeric model, and when evaluating domained interpolants and kriging estimates. Projects with more estimation objects, block models and evaluations will see the most improvements with the changes made in this area.
“This improvement reduced the processing time of an extremely large complex project that containing a significant number of domained estimation objects, from over 200 hours to well below 70 hours”
Note that the processing improvement seen in individual projects will vary, depending on the complexity and structure of each project. The biggest gains will be found in projects with many numeric estimators and/or domained estimation objects in combination with Octree block models.
For any users who have not yet tested or adopted the Octree model, we strongly recommend you try out this block model format. A huge amount of effort has gone into optimising this highly efficient model format, and the benefits in terms of both evaluation and storage will be apparent. As an example, the table below compares the project size, size on disk, number of rows and time to create/evaluate both Leapfrog’s previous sub-block model and the new Octree model, between the current release (Leapfrog 2021.1.3) and the new release (Leapfrog 2021.2). The base block model is relatively small (94,424 parent cells of 5x20x20m, and 16x8x8 subcelling), and was evaluated with 2 GM’s, a combined kriging estimator, 3 domain kriging evaluations and a grouped mesh evaluation.
These performance improvements will enable Geoscientists to focus more on interpreting result rather than spending time waiting on processing. It will also free up geoscientists to make changes further up the project tree without the fear that it will cause a significant amount of reprocessing thus allowing the customer to reach the right decision faster.
We are continually developing ways to modernise the Leapfrog application and add new functionality to ensure we are keeping up with the fast pace of digital transformation in geoscience and cloud computing.