inference only. This is an IoT/robotics play for Intel, combined with RealSense. The only feasible product at the moment for training large models is NVidia G

inference only. This is an IoT/robotics play for Intel, combined with RealSense. The only feasible product at the moment for training large models is NVidia GPUs. Maybe AMD in 1-2 years time, if they decide to wake up and invest in software optimisation for deep learning