setChunkSize {doRedis}

Set the default granularity of distributed tasks.

Description

A job is the collection of all tasks in a foreach loop. A task is a collection of loop iterations of at most size chunkSize. R workers are assigned work by task in blocks of at most chunkSize loop iterations per task. The default value is one iteration per task. Setting the default chunk size larger for shorter-running jobs can substantially improve performance. Setting this value too high can negatively impact load-balancing across workers, however.

Usage

setChunkSize(value = 1)

Arguments

value

positive integer chunk size setting

Value

value is invisibly returned; this value is called for its side effect.

Examples

# Only run if a Redis server is running
if (redux::redis_available()) {

# Start a single local R worker process
startLocalWorkers(n=1, queue="jobs", linger=1)

# Register the work queue with the coordinator R process
registerDoRedis("jobs")

# Compare verbose task submission output from...
setChunkSize(1)
foreach(j=1:4, .combine=c, .verbose=TRUE) %dopar% j

# with the verbose task submission output from:
setChunkSize(2)
foreach(j=1:4, .combine=c, .verbose=TRUE) %dopar% j

# Clean up
removeQueue("jobs")
}