MLPerf Inference Benchmarks Show NVIDIA Strength, Company Expands Jetson AI Edge Offerings

A few months back, I wrote about the MLPerf consortium and the release of its Inference v0.5 benchmark. MLPerf had previously disclosed some performance results from its Training v0.6 benchmark, but training is only part of the machine learning equation. It is when the training process is complete and weightings have been assigned to the dataset that a neural network can intelligently infer things from that data — this process is what is referred to as inference.