Figure 1.

Overview of the system. The user controls the system through a cluster of conventional computers by sending configuration and spike data to a number of modules that each carry a wafer. These wafer modules are interconnected with a high-speed network to exchange spike events. The wafer contains identical building blocks, of which one is shown in an expanded view. The proposed extension to the BrainScaleS wafer-scale system in form of the embedded plasticity processor is marked in red. Input/output access from the processor to other components of the building block is indicated with triangles.