3 ms·A Scalable Near-Memory Architecture for Training DNN on Large In-Memory Datasets2 points by lainon 8y ago