DATA PROCESSING METHOD AND APPARATUS, DEVICE, AND MEDIUM

This application provides a data processing method, applied to a computing cluster. The computing cluster includes a plurality of groups of an all to all structure, and each group includes a switching node and a computing node. The method includes: A switching node, that is, a first top node, that is in a first group and that is configured to communicate with a second top node in a second group receives first data sent by a child node; then the first top node receives second data sent by the second top node; and then the first top node aggregates the first data and the second data, to obtain third data. Because top nodes in a plurality of groups participate in aggregation communication and aggregate data, aggregation efficiency is improved. In addition, the top nodes in the plurality of groups are used to aggregate data, so that node utilization is improved, node resources are prevented from being idle, and resource utilization is improved.