Abstract
In material innovation, more computational resources are often required than are readily available when it comes to reducing time-to-market. The results are unwanted delays and elevated costs for IT acquisition and maintenance. The on-premises approach is also inflexible regarding varying simulation demands: in peak times, limited local computation resources lead to long waiting times, while in off-peak times, the expensive hardware, a depreciating asset, might sit unused.
In 2020, M2M - together with the Swiss company KaleidoSim - started to offer a Software-as-a-Service (SaaS) solution to run GeoDict simulations in the cloud. Through a simple web interface, users can upload simulation cases and run large parameter studies via batch processing in the cloud. Exploiting the practically limitless computational resources of the cloud, extremely large numbers of simulation cases can run in parallel, reducing their total runtime to the time required for a single simulation. Recently, we started offering GPU instances as well, so that the AI capabilities of GeoDict can also be fully utilized (e.g., GeoDict- AI, FiberFind-AI, ImportGeo-VOL).
M2M now offers another solution based on Amazon Web Services (AWS) for use-cases requiring interactive pre-/post-processing or visualization. Here, users start powerful virtual machine instances and connect to a remote desktop session with pre-installed GeoDict - all within 5 minutes - using only a web browser. No expensive hardware or local GeoDict installation is required.
Talking about flexibility: for KaleidoSim batch processing, M2M offers a pre-paid model where users buy virtual CPU (VCPU)-hours upfront. For the interactive AWS solution, M2M offers a pay-per-use or bringyour- own-license (BYOL) model. Between the three options of traditional on-premises installation, cloud batch processing, and interactive GeoDict sessions in the cloud, we aim to address your requirements and help you make more effective use of GeoDict and your time.
