[R-SIG-Mac] Different performance with different R versions

Ista Zahn i@t@z@hn @ending from gm@il@com
Fri Dec 14 11:58:25 CET 2018

This isn't Mac specific (I can reproduce it in Linux). Switching the
igraph from version from 1.2.2 (slow) to 0.7.1 (fast) is enough, so
this also has nothing to do with R itself. This should be reported to
https://igraph.org/r/, either on the mailing list, bug tracker, or


On Fri, Dec 14, 2018 at 3:24 AM Cowan, R (MERIT)
<r.cowan using maastrichtuniversity.nl> wrote:
> Thanks for the help.
> Attached are 3 files:
> The simulation script as simulationR-R.txt
> Rprof() output from R 3.1.3 as Rprof-3-1-3.txt
> Rprof() output from R 3.5.1 as Rprof-3-5-1.txt
> There is information about versions and system times at the top of the simulation script. I had a student try it on her windows machine. She has R version 3.4 installed. It performs the same as version 3.5, so whatever changes to R account for the change in performance probably came before version 3.4 (and suggests it is not the OS).
> I hope the script contains enough documentation that you can decipher it. There are comments about what the algorithm is doing but also comments about things I have done to try to isolate the problem (deleting stuff and moving things around, which by the way makes nonsense of the simulation itself, but that is not an issue here).
> It uses the igraph package, but I re-arranged things such that I am pretty sure that is not the problem. I realized just now I could get rid of the igraph stuff altogether for debugging purposes  if that would help. Let me know about that.
> Thanks for your help. I hope you can track it down, because this kind of speed degradation make me really hesitate to upgrade.
> Robin Cowan
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac

More information about the R-SIG-Mac mailing list