| Line 90... |
Line 90... |
| 90 |
on the \code{p} nodes. When the first job completes, the next job is
|
90 |
on the \code{p} nodes. When the first job completes, the next job is
|
| 91 |
placed on the node that has become free; this continues until all jobs
|
91 |
placed on the node that has become free; this continues until all jobs
|
| 92 |
are complete. Using \code{clusterApplyLB} can result in better
|
92 |
are complete. Using \code{clusterApplyLB} can result in better
|
| 93 |
cluster utilization than using \code{clusterApply}, but increased
|
93 |
cluster utilization than using \code{clusterApply}, but increased
|
| 94 |
communication can reduce performance. Furthermore, the node that
|
94 |
communication can reduce performance. Furthermore, the node that
|
| 95 |
executes a particular job is non-deterministic.
|
95 |
executes a particular job is non-deterministic. This means that
|
| - |
|
96 |
simulations that assign RNG streams to nodes will not be reproducible.
|
| 96 |
|
97 |
|
| 97 |
\code{clusterMap} is a multi-argument version of \code{clusterApply},
|
98 |
\code{clusterMap} is a multi-argument version of \code{clusterApply},
|
| 98 |
analogous to \code{\link{mapply}} and \code{\link{Map}}. If
|
99 |
analogous to \code{\link{mapply}} and \code{\link{Map}}. If
|
| 99 |
\code{RECYCLE} is true shorter arguments are recycled (and either none
|
100 |
\code{RECYCLE} is true shorter arguments are recycled (and either none
|
| 100 |
or all must be of length zero); otherwise, the result length is the
|
101 |
or all must be of length zero); otherwise, the result length is the
|
| Line 123... |
Line 124... |
| 123 |
\code{X} takes quite variable amounts of time, and either the function is
|
124 |
\code{X} takes quite variable amounts of time, and either the function is
|
| 124 |
deterministic or reproducible results are not required. Chunks of
|
125 |
deterministic or reproducible results are not required. Chunks of
|
| 125 |
computation are allocated dynamically to nodes using
|
126 |
computation are allocated dynamically to nodes using
|
| 126 |
\code{clusterApplyLB}. From \R 3.5.0, the default number of chunks is
|
127 |
\code{clusterApplyLB}. From \R 3.5.0, the default number of chunks is
|
| 127 |
twice the number of nodes. Before \R 3.5.0, the (fixed) number of chunks
|
128 |
twice the number of nodes. Before \R 3.5.0, the (fixed) number of chunks
|
| 128 |
was the same as the number of nodes.
|
129 |
was the same as the number of nodes. As for \code{clusterApplyLB},
|
| - |
|
130 |
with load balancing the node that executes a particular job is
|
| - |
|
131 |
non-deterministic and simulations that assign RNG streams to nodes
|
| - |
|
132 |
will not be reproducible.
|
| 129 |
|
133 |
|
| 130 |
\code{parRapply} and \code{parCapply} are parallel row and column
|
134 |
\code{parRapply} and \code{parCapply} are parallel row and column
|
| 131 |
\code{apply} functions for a matrix \code{x}; they may be slightly
|
135 |
\code{apply} functions for a matrix \code{x}; they may be slightly
|
| 132 |
more efficient than \code{parApply} but do less post-processing of the
|
136 |
more efficient than \code{parApply} but do less post-processing of the
|
| 133 |
result.
|
137 |
result.
|
The R Project SVN
R