Abstract:
The paper deals with the issues of observability and controllability of internal states of the recurrent dataflow processor. The implementation of tools that enable the developers to maximally quickly detect, isolate, and correct the error issues is considered. The most effective decisions considering architectural novelty and specificity of the recurrent dataflow processor organization are suggested.