From 140f1e6ecf9efc93ee13e3a1d8af77343200e614 Mon Sep 17 00:00:00 2001 From: giannozz Date: Fri, 12 Oct 2012 13:01:12 +0000 Subject: [PATCH] Added remark on how to reduce I/O git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@9529 c92efa57-630b-4861-b058-cf58834340f0 --- Doc/user_guide.tex | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) diff --git a/Doc/user_guide.tex b/Doc/user_guide.tex index 67752e579..1e3fac103 100644 --- a/Doc/user_guide.tex +++ b/Doc/user_guide.tex @@ -1449,9 +1449,10 @@ of processors of each pool. \paragraph{Massively parallel calculations} For very large jobs (i.e. O(1000) atoms or more) or for very long jobs, to be run on massively parallel machines (e.g. IBM BlueGene) it is -crucial to use in an effective way all available parallelization levels. Without a judicious choice of -parameters, large jobs will find a stumbling block in either memory or -CPU requirements. +crucial to use in an effective way all available parallelization levels. +Without a judicious choice of parameters, large jobs will find a +stumbling block in either memory or CPU requirements. Note that I/O +may also become a limiting factor. Since v.4.1, ScaLAPACK can be used to diagonalize block distributed matrices, yielding better speed-up than the internal algorithms for @@ -1556,7 +1557,9 @@ You can use input variable \texttt{disk\_io='minimal'}, or even into trouble (or into angry system managers) with excessive I/O with \pwx. The code will store wavefunctions into RAM during the calculation. Note however that this will increase your memory usage and may limit -or prevent restarting from interrupted runs. +or prevent restarting from interrupted runs. For very large runs, +you may also want to use \texttt{wf\_collect=.false.} and (CP only) +\texttt{saverho=.false.} to reduce I/O to the strict minimum. \subsection{Tricks and problems}