search for: parlapplylb2

...overhead can be reduced by sending all additional arguments to the workers once and for all (a trick I have seen on page 20 of "Parallel R" by QE McCallum). Here is my version of it (the only difference to McCallum's code is that I moved LB.init() and LB.worker() to the inside of parLapplyLB2()): parLapplyLB2 <- function(cl, x, fun, ...) { LB.init <- function(fun, ...) { assign(".LB.fun", fun, pos=globalenv()) assign(".LB.args", list(...), pos=globalenv()) NULL } LB.worker &lt...