bowden at stats.ox.ac.uk
2008-Jun-05 16:10 UTC
[Rd] memory.size() for large memory usage (PR#11596)
This amusing behaviour is from R2.6.1 on WinXP Pro SP2 running with boot.ini /3GB flag, exploring memory limits. =20 When actual (object, not total) memory usage hits 2048 MB, memory.size starts counting down again, but reports a negative amount, e.g. -2046.333. Is this an intended (or unavoidable) feature?) =20 Thanks, =20 Rory Bowden bowden at stats.ox.ac.uk =20 [[alternative HTML version deleted]]
ripley at stats.ox.ac.uk
2008-Jun-06 07:55 UTC
[Rd] memory.size() for large memory usage (PR#11596)
On Thu, 5 Jun 2008, bowden at stats.ox.ac.uk wrote:> This amusing behaviour is from R2.6.1 on WinXP Pro SP2 running with boot.in> i /3GB flag, exploring memory limits. > > =20 > > When actual (object, not total) memory usage hits 2048 MB, memory.size star> ts counting down again, but reports a negative amount, e.g. -2046.333. Is t> his an intended (or unavoidable) feature?)Not any more (it was a feature of the malloc used, but that has been updated and R hasn't followed.). Will be changed for 2.7.1.> > =20 > > Thanks, > > =20 > > Rory Bowden > > bowden at stats.ox.ac.uk-- 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