ccl: iend, mxcore summary

Hi,
 Thanx, to all those numerous people who responded to my question
 about what iend and mxcore are in G98 an why a job would fail if iend
 would be smaller then mxcore. Below I included the answer of Doug Fox
 (from Gaussian) which summarizes almost everything.
 Thanx,
 Martijn
 Answer:
 Gaussian 98 works with a fixed memory allocation which is set
 in the input.  The default is 6MW (48MB) but you simply need to
 add a line at the top of the input of the form
 %mem=8MW
 to increase this.
 This is related to the error message as follows, iend is the amount
 of memory that Gaussian needs at this step.  MxCore is the amount of
 the total memory left at this point.  So when you get this message
 you need to increase the %mem value overall by the difference between
 these two.
 Looking at the fragment below you have the default memory as noted by
 >  Leave Link  301 at Thu Jun  5 14:59:36 2003, MaxMem=    6291456
 cpu:
 and then some of it has been previously allocated at the time this
 failed,
 >  ShPair-CalcS2-2 allocation failure:  iend,mxcore=   4107035
 4060616
 and only 4MW are left.  The calculation was only a little short,
 needing
 4107035 - 4060616 = 46419 words but generally you will do better to
 allocate some excess.  So
 %mem=7mw
 would be plenty.
 You can increase the default if you have more memory available by
 putting a line in the /opt/g98/g98Default.Route file,
 -M- 12MW
 to request 96MB.  Just doing more memory will not always make it
 faster so don't allocate all of the available memory.  For a 512MB
 machine making the default 256MB is reasonable.  You can increase it
 for individual jobs if needed, as noted above.
 Original question:
 >
 > Hi,
 >
 > We're trying to find the limitations of one of our machines (for
 > getting the specs for our new machines) and have noticed that some
 jobs
 > crash upon increasing the basis-set with the following error
 message:
 >
 > here are  1488 symmetry adapted basis functions of A   symmetry.
 >  Crude estimate of integral set expansion from redundant
 > integrals=1.000.
 >  Integral buffers will be    262144 words long.
 >  Raffenetti 2 integral format.
 >  Two-electron integral symmetry is turned on.
 >   1488 basis functions     3232 primitive gaussians
 >    432 alpha electrons      432 beta electrons
 >        nuclear repulsion energy     32002.1321856990 Hartrees.
 >       0.1
 >  (Enter /opt/g98/g98/l302.exe)
 >  One-electron integrals computed using PRISM.
 >  Error termination via Lnk1e in /opt/g98/g98/l302.exe.
 >
 > Can somebody explain what's happening here? and what are iend and
 > mxcore precisely?
 >
 > Thanx in advance
 >
 > Cheers,
 >
 > Martijn-------------------------------------------------------------
 ------------
 Martijn Zwijnenburg
 Lab. of Applied Organic Chemistry and Catalysis
 Delft University of Technology
 Julianalaan 136
 2628 BL Delft
 The Netherlands
 Tel: 0031-(0)152782691
 Fax: 0031-(0)152784700
 e-mail: M.A.Zwijnenburg$at$tnw.tudelft.nl
 web page: http://come.to/tock