| parallel_opts {geneTree} |
R Documentation |
Override the parallelism of a trained model (the perturbation solver) or of a
Description
configuration object (the model training).
Usage
parallel_opts(tree,
threads = -1,
threads.per.model = -1);
Arguments
tree
the GeneTree model or the configuration object.
threads
the outer (per gene) degree of parallelism; <= 0 restores the automatic mode which uses all of the available processors. [as integer]
threads.per.model
the TGBoost thread number of a single gene model: -1 automatic (1 when the genes are trained in parallel), 0 sequential, >0 explicit. It only affects the model training, it is ignored for the perturbation solving. [as integer]
env
the R# runtime environment object. [as Environment]
Details
the training is parallel over the target genes, therefore the per model threads are resolved to 1 automatically: nested parallelism (outer over genes plus inner tree building) starves the .NET thread pool and the CPU utilization collapses to a few cores even on a 128 threads server.
Authors
biosystem
Value
the input object with the new parallelism settings.
clr value class
Examples
[Package
geneTree version 1.0.0.0
Index]