Abstract: A method and a protective apparatus for protecting a laser device are provided. By connecting the laser device to a protective housing to safeguard laser engraving operations, safety during a laser engraving process is enhanced. Furthermore, a support plate of the protective apparatus is capable of vertically moving, thereby enabling the protective housing to adapt to objects of varying heights and different types of the laser device to consistently achieve laser engraving within an optimal engraving focal range, an applicability scope of the protective apparatus is expanded, and user experience is further improved. Additionally, two guide structures are respectively disposed a left side of the support plate and a right side of the support plate to drive the support plate to vertically move, thereby ensuring stable vertical movement during the laser engraving process, yielding superior engraving results with enhanced precision.

Abstract: A quantization method for a neural network model is provided. The quantization method includes: determining sensitivities corresponding to one candidate max weight error (MWE) among candidate MWEs corresponding to the target layer, the sensitivities sensitivity of the neural network model to quantization; determining a target MWE corresponding to the target layer, based on the sensitivities; and based on the determined target MWE, quantizing weights included in the target layer from a first data format to a second data format.

Abstract: A precision allocation method and device based on a neural network processor are provided. The precision allocation method includes allocating a weight of a neural network to a multiplier column of a neural network processor, determining a lower tolerance for the multiplier column, and selecting a first data type for the multiplier column from a plurality of data types based on the lower tolerance, wherein each of the plurality of data types corresponds to a different precision level, and performing, by the neural network processor, a multiplication operation based on the weight and the first data type.

Abstract: In order to solve the technical problem that in the current real-time distributed systems such as multi-antenna MIMO system, the implementation of synchronization between distributed modules by means of software is of a large time overhead, the invention proposes a synchronization method and device for the real-time distributed system, the distributed module determines if the present module is able to operate synchronously, at a synchronization start time that is unified with the other distributed modules; sets the present module's operate indication signal in a logic circuit, based on the determining result; and decides whether the present module is to operate synchronously, following a rule unified with the other distributed modules, based on the present module's operate indication signal along with the operate indication signals in the logic circuit set by the other modules.

Abstract: In order to solve the technical problem that in the current real-time distributed systems such as multi-antenna MIMO system, the implementation of synchronization between distributed modules by means of software is of a large time overhead, the invention proposes a synchronization method and device for the real-time distributed system, the distributed module determines if the present module is able to operate synchronously, at a synchronization start time that is unified with the other distributed modules; sets the present module's operate indication signal in a logic circuit, based on the determining result; and decides whether the present module is to operate synchronously, following a rule unified with the other distributed modules, based on the present module's operate indication signal along with the operate indication signals in the logic circuit set by the other modules.