I ran some sample code that others showed a 10x increase when using MRO as opposed to CRAN, and I saw nothing significant. Ok, maybe, it was statistically significant, but the variance between runs was as much as the improvement.
Did CRAN 3.6 do something to improve, or is it possible the MRO version is not set up correctly. The results I obtained were both similar to the faster results for a simple operation; however I likely have faster computer than some even though it is a bit dated I7-6800 I repeated this within RStudio and JUPYTER notebook. Note, it was slight faster in RStudio.