Earlier this year, we talked about performance breakthroughs enabled by two new technologies we developed. Related to data augmentation and tensor parameter interleaving, these technologies enabled a single Tesla V100 GPU to do a ResNet-50 training run in just under 24 hours (1,350 images/second). We also discussed how these same technologies enabled a single DGX-1 server

The post NVIDIA DGX-2 AI Supercomputer and New Tesla T4 Set Image Recognition Records for Training and Inference appeared first on NVIDIA Developer News Center.