Cache-Aware Graph Processing for High-Throughput GNN Training. ijaies 2026, 1 (1).