[Rd] Re: [R] Running R from CD?

[Prof Brian Ripley]

On Sat, 4 Dec 2004, M. Edward Borasky wrote:

> On Sat, 2004-12-04 at 19:47 +0000, Prof Brian Ripley wrote:
>> This is now solved in R-devel: it runs in the slowest scenario below with
>> a startup time of about 2.5secs and no appreciable delay when running.
>
> I've never had any performance issues in R (or any other packages,
> really) with Dirk Eddelbuettel's Quantian "live" CDs or DVDs, and they
> run the whole OS -- not just the applications -- from a *compressed*
> CD/DVD. The key on Linux is, I think, to have plenty of RAM for I/O
> buffering in addition to what's required for your R datasets -- my
> smallest machine is a 256 MB 933 MHz PIII. I can't comment on the
> Windows kernel, though -- I do Linux performance engineering for a
> living :).

[The part omitted here said, supported by data, that other people did have 
performance issues. In particular the `slowest scenario' was not running 
from a CD, but a USB drive.  Please do not remove the essential parts of 
other people's threads without indication -- it is discourteous to both 
poster and the reader.]

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595