Clarification On data.table::getDTthreads(verbose = TRUE)

[drag05]

Argument definition in setDTthreads() function documentation seems to confuse the statements of getDTthreads(verbose = TRUE:

> data.table::getDTthreads(verbose = TRUE)
  OpenMP version (_OPENMP)       201511
  omp_get_num_procs()            16
  R_DATATABLE_NUM_PROCS_PERCENT  unset (default 50)
  R_DATATABLE_NUM_THREADS                 unset
  R_DATATABLE_THROTTLE                         unset (default 1024)
  omp_get_thread_limit()                                 2147483647
  omp_get_max_threads()                               16
  OMP_THREAD_LIMIT                                 unset
  OMP_NUM_THREADS                               unset
  RestoreAfterFork                                         true
  data.table is using 8 threads with throttle==1024. See ?setDTthreads.
[1] 8

Here, argument "threads" refers to CPU logical cores. As the "percent" default is 50, the value is 8 as stated by next-to-the-last-row in above code. However the env. variable R_DATATABLE_NUM_THREADS is "unset" leading to confusion. This function documentation should be clarified further. Thank you!

[ben-schwen]

I don't see how the documentation is wrong.

Sys.getenv("R_DATATABLE_NUM_PROCS_PERCENT") will return "", same as Sys.getenv("R_DATATABLE_NUM_THREADS") when not set.

[drag05]

@ben-schwen Documentation is not wrong. Here is why I think it only needs to be expounded:

1. Argument "threads" in setDTthreads refers to logical CPU cores (hardware) while the generic "threads" refers to software processes. Usually, there are hundreds or thousands of threads involved when running programs on CPU with 18, 24 etc. CPU logical cores. Hence, the confusion;
2. There are two groups of environment variables in above example: the "R_DATATABLE" group and the "OMP" group. Expound on advantages and/or probable conflicts when setting values in both groups simultaneously. Hence, the need for documentation clarification.
   Thank you!