5results about How to "Reduce storage load" patented technology

The invention provides a recording and broadcasting equipment control method, system and device, a storage medium and electronic equipment, and the control method comprises the steps: obtaining target equipment data of target recording and broadcasting equipment; generating a target operation page data request according to the target equipment data, and sending the target operation page data request to a server; receiving target operation page data issued by the server, and displaying a target operation page according to the target operation page data; and in response to a trigger operation of a user on the target operation page, obtaining a control instruction corresponding to the trigger operation, and sending the control instruction to the target recording and broadcasting device. According to the invention, the use cost of the recording and broadcasting equipment can be reduced, the occupied memory resources of the terminal equipment are saved, and the load of the memory resources of the terminal equipment is reduced.

The invention discloses a block chain node data exchange method, a readable storage medium and equipment, and the method comprises the steps: carrying out the coding processing of original block data, and obtaining a plurality of block units; and calculating the storage load difference, the block decoding communication overhead and the node reputation of the target block chain node group of each block unit, and determining an optimal block unit allocation scheme with low load difference and low decoding communication overhead among the target block chain nodes. According to the invention, based on the storage load difference, the block decoding communication overhead and the node reputation of the target block chain node group, the optimal block unit allocation scheme of low load difference and low decoding communication overhead between the nodes is determined, and the block units are stored according to the optimal block unit allocation scheme, so that the storage load of the nodes is greatly reduced, and the storage efficiency of the nodes is improved. The problems of unreasonable block unit distribution and overlarge node storage pressure in the current block chain are solved.

The application provides a direct current offset correction circuit based on a lookup table, a zero intermediate frequency receiver and a direct current offset correction method based on a lookup table. The direct current offset correction circuit based on a lookup table comprises a storage module, an analog coarse correction module and a digital fine correction module. In the receiving channel of the zero intermediate frequency receiver, the analog coarse correction module is used to perform analog coarse correction on the direct current offset in the analog signal, and the digital fine correction module is used to perform digital fine correction on the direct current offset in the digital signal. The analog intermediate frequency correction and the digital baseband correction are considered, and the correction precision of the direct current offset is improved. Meanwhile, the direct current offset correction lookup tables generated by radio frequency and intermediate frequency devices are separately established. During correction, indexing and operation are performed based on the analog coarse correction module and the digital fine correction module. Only a small number of calculation input correction parameters need to be saved, the total capacity of the lookup table is reduced, and the storage load of the storage module is reduced.

The present invention discloses a memory scheduling method for pipeline parallel training, including: deployment phase: segmenting a neural network model with network layers as units to form different network partitions; deploying network partitions on different computing nodes; training phase: each computing node is only responsible for the calculation of the deployed network partition; each computing node communicates data with upstream and downstream computing nodes; data transfer mechanism: unloading activation data in the computing node to the CPU memory, waiting for the data to be retrieved when it needs to be used again; data processing phase: compressing the activation data and then sending it to the downstream computing node or CPU memory; after the downstream computing node receives the activation data, or after the original computing node retrieves the activation data from the CPU memory, the computing node decompresses the activation data once. The present invention has the advantages of simple principle, easy implementation, and can reduce storage load and communication overhead.

Embodiments of the present application disclose a data storage method, related device and equipment, which are used for avoiding resource waste of data storage. The method comprises the following steps: obtaining to-be-transmitted data, performing fragmentation processing on the to-be-transmitted data according to a target data type to obtain K pieces of fragmented data and an actual storage amount corresponding to each piece of fragmented data, allocating a preset storage amount to each piece of fragmented data according to the actual storage amount corresponding to each piece of fragmented data, determining a storage address of each piece of fragmented data if a data tag of each piece of fragmented data exists in a historical storage record, the historical storage record comprising a mapping relationship between data tags and storage addresses within a historical time, sending a resource creation request to a server according to the storage address of each piece of fragmented data, so that the server creates storage resources matching the preset storage amount for each piece of fragmented data, and sending the K pieces of fragmented data to the server, so that the server stores the K pieces of fragmented data.