- Reduced Memory Footprint
- The current resource manager accounts for a significant portion of the AI Engine driver's memory usage, particularly on devices with limited memory capacity.
- Enhanced Memory Management
- Dynamic memory allocation requirements associated with the resource manager hinder compliance with stringent automotive code safety standards.
- Streamlined Resource Requests
- Given that users primarily interact with the FAL layer to request resources, bypassing direct interaction with the resource manager APIs, the new implementation aims to streamline resource management by empowering the FAL to handle resource requests directly. This eliminates unnecessary calls to the user space driver, resulting in improved efficiency and resource utilization.
These architectural changes seek to optimize the performance and resource utilization of the AI Engine driver, ensuring smoother operation across different hardware configurations while adhering to industry-standard safety guidelines.