Skip to contents

Building a pyramid runs in parallel. a5pyramid_threads() reports the number of threads builds use by default, and a5pyramid_threads(n) sets it for the session. As with options() and a5R::a5_threads(), setting returns the previous value, so with(a5pyramid_threads(2), expr) runs expr on two threads and then restores the setting, even if expr fails. The threads argument of a5_pyramid() and a5_write_pyramid() overrides it for one call.

Usage

a5pyramid_threads(n = NULL)

# S3 method for class 'a5pyramid_threads'
with(data, expr, ...)

Arguments

n

NULL (the default) to report the current count, an integer of at least 1 to set it, or the value of an earlier a5pyramid_threads(n) to restore that setting.

data

The value of a5pyramid_threads(n).

expr

Code to run with that thread count.

...

Unused.

Value

a5pyramid_threads(): the current count, an integer. a5pyramid_threads(n): the previous setting, invisibly, as an a5pyramid_threads object. with(): the value of expr.

Details

The default is resolved in this order: the option a5pyramid.threads, the environment variable A5PYRAMID_THREADS, then every CPU available to the R process (which respects affinity masks and container CPU quotas). The count is independent of a5R::a5_threads(). Results do not depend on it.

Examples

a5pyramid_threads()
#> [1] 4
old <- a5pyramid_threads(2)
a5pyramid_threads()
#> [1] 2
a5pyramid_threads(old) # restore

with(a5pyramid_threads(1), a5pyramid_threads())
#> [1] 1