Abstract: Memory management is provided for processors, such as GPUs used to process data using a trained machine learning model. Requests received to a CPU can be stored to a request queue until the queue is full, or until a timeout value has been reached for periods of lower activity. The requests can then be batched and sent to a GPU as a single message on a single thread. Memory can be pre-allocated, and the trained model loaded into GPU memory once for processing of the relevant batches. The individual requests can be processed by the GPU and the results analyzed to determine at least a subset of results to return to the CPU, which can be provided back as results of the processing.

Abstract: Systems and methods for rendering 3D maps may highlight a feature in a 3D map while preserving depth. A map tool of a mapping or navigation application that detects the selection of a feature in a 3D map (e.g., by touch) may perform a ray intersection to determine the feature that was selected. The map tool may capture the frame to be displayed (with the selected feature highlighted) in several steps. Each step may translate the map about a pivot point of the selected map feature (e.g., in three or four directions) to capture a new frame. The captured frames may be blended together to create a blurred map view that depicts 3D depth in the scene. A crisp version of the selected feature may then be rendered within the otherwise blurred 3D map. Color, brightness, contrast, or saturation values may be modified to further highlight the selected feature.

Abstract: Methods and systems are provided for efficiently identifying map tiles of a raised-relief map to retrieve from a server. An electronic device can use estimates of height(s) for various region(s) of the map to determine map tiles that are likely viewable from a given position of a virtual camera. The device can calculate the intersection of the field of view of the virtual camera with the estimated heights to determine a location of the map tiles (e.g., as determined by a 2D grid) needed. In this manner, the electronic device can retrieve, from a map server, the map tiles needed to display the image, without retrieving extraneous tiles that are not needed. Identifying such tiles can reduce the amount of data to be sent across a network and reduce the number of requests for tiles, since the correct tiles can be obtained with the first request.

Abstract: Systems and methods for rendering 3D maps may highlight a feature in a 3D map while preserving depth. A map tool of a mapping or navigation application that detects the selection of a feature in a 3D map (e.g., by touch) may perform a ray intersection to determine the feature that was selected. The map tool may capture the frame to be displayed (with the selected feature highlighted) in several steps. Each step may translate the map about a pivot point of the selected map feature (e.g., in three or four directions) to capture a new frame. The captured frames may be blended together to create a blurred map view that depicts 3D depth in the scene. A crisp version of the selected feature may then be rendered within the otherwise blurred 3D map. Color, brightness, contrast, or saturation values may be modified to further highlight the selected feature.

Abstract: Methods and systems are provided for efficiently identifying map tiles of a raised-relief map to retrieve from a server. An electronic device can use estimates of height(s) for various region(s) of the map to determine map tiles that are likely viewable from a given position of a virtual camera. The device can calculate the intersection of the field of view of the virtual camera with the estimated heights to determine a location of the map tiles (e.g., as determined by a 2D grid) needed. In this manner, the electronic device can retrieve, from a map server, the map tiles needed to display the image, without retrieving extraneous tiles that are not needed. Identifying such tiles can reduce the amount of data to be sent across a network and reduce the number of requests for tiles, since the correct tiles can be obtained with the first request.