A lot hangs for Nvidia on the success of its latest Tesla T4 AI chip, which certainly boasts impressive specifications – even if we only have the company’s word to take for some of the more specific performance claims. We can all agree it is almost 12 times faster than Nvidia’s older P4 chip, at the half precision FP16 floating point arithmetic relevant for AI calculations, at 65 teraflops against 5.5 teraflops, with a similar gain for integer arithmetic. But just as important are the associated new tools and software packages for development and optimization of AI and machine learning models, with a focus on video. It is no surprise that one of Nvidia’s headline datapoints is the ability of…