Skip to content

5 Step X: Performance parameters

Torsten Pook edited this page Mar 20, 2019 · 1 revision

Parameters: recoding, recoding_notneeded, fast_compiler, intersect_func, c_dhm_mode, parallel_window, window_overlap, window_cores

To reduce the computation time there are some additional possible options. Internal computations are faster when a low number of different characters is use. To change the coding to major_variant “A”, minor_variant “C” in every SNP set the parameter recoding to TRUE. If this is already done you can further use recoding_notneeded to skip this recoding step and still profit from the advantages of the recoding.

To further increase computation time one can active parallel computing via parallel_window. Here the dataset is split into windows containings parallel_window markers. On defaults window do not overlap but can via window_overlap. The number of cores to be used is controlled via window_cores.

All other options are there mostly because of testing purposes and should already be set to the optimal value. Fast_compiler enables the compiler-packages and just-in-time computing. Intersect_func loads in a more efficient variant of base::intersect and c_dhm_mode uses bit-wise-computing of the dataset.

Clone this wiki locally