Tutorial - Mapping


TTPython compiles to graphs to represent the parallelism expected of a distributed application. In many cases, the programmer will need specific SQs to run on a particular hardware element or ensemble. This mapping constraint could be used to restrict a sampling SQ to run on an ensemble that has a temperature sensor or an image processing SQ to be run on an emsemble that has a GPU. Mapping constraints may aso dictate end-to-end requirements, like a maximal latency between a series of SQs. This tutorial will show you how to specify mapping constraints and how you can expect them to be satisfied as the graph is distributed among the available ensembles in the system.

** This is still WIP on syntax and semantics **