Here is a very simple and probably inefficient way to do something like that. I only ran the code for three seconds and got 3.1 million values occupying 24Mb. If you generate the random numbers in blocks bigger than 10000, you will probably get even more values.
Out of curiosity, why do you want to do this?
Start <- Sys.time()
OUT <- vector(mode = "numeric")
while (Sys.time() - Start < 3) { #run for 3 seconds
OUT <- c(OUT, runif(10000, 0, 10))
}