The R script of my previous post was much too long and had a mistake in adding a row to the data.frame (mixing character and numeric data in a vector). I have also added the quantile value to the table to check (column result).

RNGmethod ctype result user system elapsed

1 Inversion loop 0.05180068 1.681 0.006 1.687

2 Inversion iteration 0.05211462 1.648 0.206 1.853

3 Kinderman-Ramage loop 0.05228179 1.105 0.061 1.166

4 Kinderman-Ramage iteration 0.05176681 1.027 0.001 1.028

5 Box-Muller loop 0.05224984 1.513 0.018 1.531

6 Box-Muller iteration 0.05208642 1.417 0.005 1.423

7 Ahrens-Dieter loop 0.05237948 1.076 0.021 1.097

8 Ahrens-Dieter iteration 0.05192184 0.993 0.000 0.993

The Ahrens-Dieter method is still the quickest.

resdata <- data.frame(RNGmethod="",ctype="", result=0, user=0, system=0, elapsed=0,stringsAsFactors=FALSE,row.names=NULL)

nlen <- 1000
iterations <- 10000
averages <- numeric(iterations)
whichp <- 0.95

for( Rmethod in c("Inversion","Kinderman-Ramage", "Box-Muller", "Ahrens-Dieter") ) {
   RNGkind(normal.kind=Rmethod)

   t <- system.time( {
   for(k in 1:iterations) { v <- rnorm(nlen); averages[k] <- mean(v) }
   result <- quantile(averages,probs=c(whichp))
   })

   resdata <- rbind(resdata, list(RNGmethod=Rmethod, ctype="loop", result=result, user=t[1],system=t[2],elapsed=t[3]))

   t <- system.time({
   averages2 <- colMeans(matrix(data=rnorm(nlen*iterations), nrow=nlen, ncol=iterations))
   result <- quantile(averages2,probs=c(whichp))
   })

   resdata <- rbind(resdata, list(RNGmethod=Rmethod, ctype="iteration", result=result, user=t[1],system=t[2],elapsed=t[3]))
}

# delete dummy first row
resdata <- resdata[2:nrow(resdata),]
row.names(resdata) <- NULL
resdata