ParallelParam-classes: Classes to configure parallel evaluation
In Streamer: Enabling stream processing of large files

Description Usage Arguments Constructors Methods Internal Class Fields and Methods Author(s) See Also Examples

Configure and register parallel calculations, e.g., for Team evaluation.

1
2
3

MulticoreParam(size = getOption("mc.cores", 2L),
    mc.set.seed = TRUE, ...)
register(param)

`size`	The number of members in the parallel cluster.
`mc.set.seed`	`logical(1)`; see `?mcparallel` on unix platforms.
`param`	A `ParallelParam` instance, such as generated by `MulticoreParam()`.
`...`	Additional arguments, e.g., `verbose`, passed to the `Streamer` class.

Use MulticoreParam to construct instances of this class.

register: Invoked with an argument param stores the param for use in subsequent parallel computation. Use NULL to clear the register. The function returns, invisibly, the previously registered parameter instance, if any.

Internal fields of this class are are described with, e.g., getRefClass("MulticoreParam")$fields.

Internal methods of this class are described with getRefClass("MulticoreParam")$methods() and getRefClass("MulticoreParam")$help().

Martin Morgan mtmorgan@fhcrc.org

Team to apply one function in parallel, DAGTeam to evaluate functions whose dependencies are represented as directed acyclic graphs.

if (.Platform$OS.type != "windows") {
    oparam <- register()       ## previous setting
    param <- MulticoreParam()  ## default multicore settings
    register(param)            ## register for future use, e.g,. Team
    register(oparam)           ## reset original
}