From a single device to distributed solutions.
As can be seen from the diagram, all domains are deployed on the same device. This implies that all cognitive load is handled by the devices resources, GPU, VRAM, CPU etc.
In a single device deployment the throughput of the solution will be dictated entirely by the device resources.
The diagram shows the same solution deployed across two devices. The throughput of the solution will be dictated by the resources available on both devices.
Devices can have differing resources. In such situations domain placement is calculated to balance cognitive load to a devices resources.
A Dabnis AI solution has the flexibility to contain multiple devices/U.C.E. instances within a department.
U.C.E. instances can be configured as: Masters; Slaves; Cascading, A hybrid node that acts as a primary to its downstream nodes while simultaneously acting as a slave to an upstream root master. This allows for distributed cognitive load, as well as a distributed solution application that can mimic, clone many types of business structures.
Does self-hosted, autonomous AI require massive data centres, their associated power plants, and drain on local water resources?
See full video