A data processing method, system, device and readable storage medium

Abstract

The present application discloses a data processing method, system, device and readable storage medium in the field of computer technology. This application connects the host and the hardware computing platform through the CXL protocol, so the host and the hardware computing platform can share each other's memory, IO and cache, so the training data does not need to be transmitted through storage media such as host memory, GPU cache, GPU memory, etc. The training data in the host memory is directly read by the hardware computing platform, thereby reducing the data transmission overhead. At the same time, the hardware computing platform can adjust the learning rate based on the current value of the moment sliding average and then calculate the new parameters of the model, so as to stabilize the model parameters, ensure the model accuracy, and improve the training efficiency. It can be seen that this scheme can reduce the data transmission overhead between the host and the hardware module and improve the model training efficiency. Correspondingly, a data processing system, device and readable storage medium provided by the present application also have the above technical effects.

Description

technical field [0001] The present application relates to the field of computer technology, and in particular, to a data processing method, system, device, and readable storage medium. Background technique [0002] Currently, model training can be done with the help of hardware modules such as GPUs. For example, the server serving as the host sends a large amount of training data to the hardware module, and the hardware module processes the training data for model training. After the model training is completed, the hardware module feeds back the trained model to the host. Due to the large amount of training data, and the data transmission between the host and the hardware module needs to pass through storage media such as host memory, GPU cache, GPU memory, etc., the data transmission overhead between the host and the hardware module is large, which will affect the model. training efficiency. [0003] Therefore, how to reduce the data transmission overhead between the hos...

AI Technical Summary

Problems solved by technology

Due to the large amount of training data, and the data transmission between the host and hardware modules needs to go through storage media such as host memory, GPU cache, and GPU memory, the data transmission overhead between the host and hardware modules is relatively large, which will affect the model. training efficiency

Images