Abstract: A ring buffer storage method includes generating data of a first output according to Q input tokens of a large language model (LLM), and writing the data of the first output into last Q column vectors of an updated first cache tensor buffer matrix. A starting memory address of a first cache tensor buffer is shifted according to the number Q of input tokens of the LLM for updating the first cache tensor buffer. The first cache tensor buffer forms a first cache tensor buffer matrix. The updated first cache tensor buffer forms the updated first cache tensor buffer matrix. The first cache tensor buffer matrix includes a plurality of space segments. Each row of the first cache tensor buffer matrix includes C space segments. C is a cache size. The plurality of space segments have continuous memory addresses.

Abstract: A method for performing camera modality adaptation in a simultaneous localization and mapping (SLAM) device is provided. The SLAM device includes a camera sensor and a SLAM processor. The method includes acquiring data from the SLAM device, and determining, based on the acquired data, an operational condition of the SLAM device. The method also includes deciding, based on the determined operational condition, a camera modality for the SLAM device. The method further includes controlling, based on the decided camera modality, a camera modality of an image sequence inputted into the SLAM processor.

Abstract: A method for performing frame rate adaptation in a simultaneous localization and mapping (SLAM) device is provided. The SLAM device includes a SLAM processor. The method includes: acquiring data from the SLAM device; determining, based on the acquired data, an operative condition of the SLAM device; deciding, based on the determined operative condition, a target frame rate for the SLAM device; and controlling, based on the decided target frame rate, a frame rate of an image sequence inputted into the SLAM processor.