SPACE4AI-D tool tackles the design space exploration of AI applications and is responsible for selecting optimal resources in each computational layer component deployment configurations and to assign components to resources while minimising the execution cost and guaranteeing the performance constraints.
SPACE4AI-D tool is Open Source. Source code is available on the AI-SPRINT GitLab repository (https://gitlab.polimi.it/ai-sprint/space4ai-d) under Apache 2.0 license.
SPACE4AI-D relies on the Performance Models component to estimate the execution time of AI application inference tasks and will be based on the open source Solid library (with MIT licence available at https://github.com/100/Solid), for the implementation of advanced meta-heuristic methods (e.g., local search, tabu-search, etc.).