The UltraMap v3 processing pipeline is a highly scalable processing system that adapts flexibly to the IT environment.
Key features and requirements are:
- Use of mid-to-high-end standard Windows PCs is possible, allowing the use of existing infrastructure
- Distributed processing in heterogeneous networks with automated load balancing ensure optimal usage of resources
- Time guided machine allows to use of dedicated machines at dedicated time slots (e.g. using a workstation for UltraMap processing duringnight time).
- Licensing scheme allows a wide spectrum of throughput needs. UltraMap can literally be executed on a laptop as well as on a processing system consisting of tens, or dozens or even hundreds of CPU cores and, as an option, additional GPU cores
- Licensing scheme supports parallel setup of small field processing hubs (e.g. for immediate on-site quality checks after a flight) as well as setup of small, medium and large processing centers
- GPU nodes deliver high-speed ups as the dense matching is ideal for a SIMD architecture such as graphics cards. Usage of CPU versus GPU can be configured to balance throughput
- New v3 machines (resource intensive machines) provide high performance—an entire machine can be used to work on one task at a time. v3 machines can either be configured as CPU only or as GPU-enabled nodes.
An UltraMap processing system usually consists of one or several front-end machines that are used to interact with the data and are not designed for processing. In addition, one or multiple processing machines are connected to the front end machine(s) and the data server(s). The processing machines handle data processing and may consist of multiple CPU and/or GPU nodes. The servers host the input, intermediate and final data. A very important part of the processing environment is the network required to transfer the data efficiently between front end, processing machines and servers.