This extension is very usefull for benchmarking models at different precisions quickly, I use it quite a lot.
What is missing though for this kind of comparisons is a way to get the standard deviation is execution time and VRAM usage in addition to the average.
This could be added in the Node Analytics tab and in the Latest Workflow Run stats too.