2results about How to "Improve data exchange efficiency" patented technology

The application relates to a signal modulation method, a demodulation method and a LoRa communication system. A total bit stream is divided into a plurality of parts, and is mapped into a frequency index set through G frequency index values. G LoRa signals are generated according to corresponding spreading factors, and are superimposed to obtain a single multiplexing signal of the spreading factor. The single multiplexing signals of all the spreading factors are superimposed to obtain a transmission signal, and signal modulation is completed. The transmission signal is received. For each single multiplexing signal, a de-chirp operation and a discrete Fourier transform are sequentially performed to obtain a frequency domain signal of each spreading factor. G peak values with the largest amplitudes in each frequency domain signal are detected, and corresponding frequency index values are recorded. The frequency index set corresponding to each spreading factor is estimated according to the frequency index values, and the bit stream in the frequency index set is recovered through reverse mapping. All the recovered bit streams are combined to obtain a total bit stream, and signal demodulation is completed. The application maximizes the transmission rate while ensuring the reliability of LoRa communication.

The invention provides a GPU data access method and system based on video memory cache, and the method comprises the steps: firstly querying whether an effective cache version of data needing to be accessed exists in a video memory, and if the effective cache version of the data needing to be accessed exists in the video memory, directly obtaining the data from the video memory by a GPU to avoid repeated access to a main memory; only when the required data are not cached in the video memory, the required data are obtained from the main memory and are sorted and loaded according to needs, so that the data exchange efficiency between the CPU and the GPU is improved. According to the method, data transmission delay caused by data redundancy is effectively reduced through a data scheduling and caching strategy, the method is particularly suitable for application scenes with high data reusability, and the overall efficiency of GPU data access is improved. According to the method, an existing hardware architecture does not need to be changed, and upper-layer application program codes do not need to be modified; by optimizing the data transmission process in the driving layer, the performance bottleneck is effectively solved, and the overall performance of the system is improved to the maximum extent under the set condition.