Re: [R] predicting memory usage

2008-02-18 Thread Moshe Olshansky
If your numbers are not integers then they are double precision which means 8 bytes per number. In such a case you will need 86000*2500*8 = 1.7Gb of memory (plus a small amount for book keeping). You will need more (and sometimes much more) if you intend to do some operations on your arrays). So yo

Re: [R] predicting memory usage

2008-02-18 Thread jim holtman
If this is numeric, then for just storing one copy, you will require 86000 * 2500 * 8 = 1.7GB of memory. You should have 3-4X that if you want to analyze it, so you might need about 6GB of physical memory and a 64-bit version of R. Is there some other alternative? Do you need all the values at o

[R] predicting memory usage

2008-02-18 Thread Federico Calboli
Hi All, is there a way of predicting memory usage? I need to build an array of 86000 by 2500 numbers (or I might create a list of 2 by 2500 arrays 43000 long). How much memory should I expect to use/need? Cheers, Fede -- Federico C. F. Calboli Department of Epidemiology and Public Health I