R vs. Stata benchmark
Uses https://github.com/prehensilecode/benchmark-stata-r/tree/slurm modified from the original by Matthieu Gomez.
Details
- R uses the packages data.table, fst, and fixest
- Stata uses the gtools, ftools, and autorename modules
Generate data sets
Generate data sets with the R script: 1-generate-datasets.r
Modify benchmark code
Need to set number of threads/cores for both the R and Stata code.
In 2-benchmark-r.r:
# setting options -- mc.cores may not be necessary: R should automatically use all cores that it can see
options(mc.cores=48)
setFixest_nthreads(48)
In 3-benchmark-stata.do:
/* benchmark */
set processors 48
Picotte Results
Bars to the right indicate that R is faster than Stata. Bars to the left indicate that Stata is faster than R.
