A groundbreaking research paper from Professor Sheng Kai's team at Xidian University has been selected for presentation at the esteemed International Conference on Supercomputing (ICS) 2025. The paper, titled "Cherry: Breaking the GPU Memory Wall for Large-Scale GNN Training via Micro-Batching," is authored by Wang Yan, a talented undergraduate student from the class of 2021, under the mentorship of team leader He Xin, who serves as the corresponding author. Wang Yan is listed as the first author of the paper. The paper tackles the formidable challenge of the GPU memory wall encountered in large-scale Graph Neural Network (GNN) training by introducing a novel micro-batch training technique named Cherry. This innovative approach integrates partition technology, augmented by message passing flow graphs, with a data loading mechanism centered on micro-batches. This fusion effectively mitigates redundancy and load imbalance within micro-batch partitions, while significantly reducing data preparation overhead during the training process. Experimental results demonstrate that Cherry not only enables large-scale GNN training beyond the constraints of device memory capacity but also surpasses existing methods in minimizing memory consumption and enhancing training efficiency. This advancement holds the potential to substantially lower hardware deployment costs for large-scale GNN training. The acceptance of this paper signifies the international acknowledgment of Professor Sheng Kai's team's pioneering research in the realm of high-performance computing. The conference will convene in Salt Lake City, Utah, USA, in June 2025.