Dear Grant,<br><br>There is enough space. The filesystem is Lustre with more than 10 TB
free. If it's not another MOLPRO limit, there is either a maximum
filesize in Lustre, or in the GA (it seems to be a GA shared file where
it happens).<br><br>Best wishes,<br>Leo<br><br><div class="gmail_quote">On 13 February 2012 16:44, Grant Hill <span dir="ltr"><<a href="mailto:Grant.Hill@glasgow.ac.uk">Grant.Hill@glasgow.ac.uk</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Dear Leo,<br>
<br>
This looks like you might be running out of disk space, would that make sense?<br>
<span class="HOEnZb"><font color="#888888"><br>
<br>
Grant<br>
</font></span><div class="HOEnZb"><div class="h5"><br>
<br>
On 13 Feb 2012, at 15:14, Leo Holroyd wrote:<br>
<br>
> Gerald and Grant, thanks for your advice. Applying F3EXTSIZE=200.0 leads to a different error, occuring just afterwards (sorry, it's rather long):<br>
><br>
> File size for 3-external integrals using MO/center triples: 113.307 GB<br>
> Not enough memory for triples, at least 31633288 Words more are needed!<br>
> Trying low memory option<br>
><br>
> Minimum memory for triples: 83436406 words<br>
><br>
> Used memory for triples: 197895686 words<br>
> caching ratio 3-ext integrals: 0.5%<br>
><br>
> AO-basis loaded. Number of functions: 920<br>
><br>
> Fitting-basis DFMP loaded for 0-2-external integrals.<br>
><br>
> Number of fitting functions: 2064<br>
> CPU time for 2-ext exchange integrals 109.71 sec, Elapsed time= 685.82 sec<br>
> CPU time for 2-ext Coulomb integrals 299.55 sec, Elapsed time= 435.60 sec<br>
> CPU time for DF-2ext transformation (total) 409.26 sec, Elapsed time= 1121.44 sec<br>
> Local fitting disabled for 3-ext integrals<br>
><br>
> Fitting-basis DFMP loaded for 3-external integrals.<br>
><br>
> Number of fitting functions: 2064<br>
><br>
> ERROR WRITING 32768 WORDS AT OFFSET 6438528226. TO FILE 5 IMPLEMENTATION=sf FILE HANDLE= -2997 IERR= -1999<br>
> ? Error<br>
> ? I/O error<br>
> ? The problem occurs in writew<br>
><br>
> GLOBAL ERROR fehler on processor 0<br>
><br>
> Does this mean that the 3-external problem has been solved and the error occurs at the next step, or is it still due to the 3-ext size limit?<br>
><br>
> This is a df-LCCSD(T)/aug-cc-pVTZ energy calculation on a stacked uracil dimer.<br>
><br>
> Many thanks,<br>
> Leo<br>
><br>
> On 2 February 2012 21:17, Gerald Knizia <<a href="mailto:knizia@theochem.uni-stuttgart.de">knizia@theochem.uni-stuttgart.de</a>> wrote:<br>
> On Thu, 2012-02-02 at 15:40 +0000, Leo Holroyd wrote:<br>
> > How does one use the F3EXTSIZE option to set the maximum file size for<br>
> > 3-external integrals? This is the error message when the size limit<br>
> > is exceeded:<br>
> ><br>
> > File size for 3-external integrals using MO/center triples: 113.307<br>
> > GB<br>
><br>
> You can use either this:<br>
> cfit,F3EXTSIZE=114.0<br>
> df-rhf<br>
> df-lccsd(t)<br>
><br>
> or this:<br>
> df-rhf<br>
> {df-lccsd(t),F3EXTSIZE=114.0}<br>
><br>
> Both should do the trick (I don't know why it is has a maximum on the<br>
> size, though).<br>
> --<br>
> Gerald Knizia<br>
><br>
><br>
><br>
</div></div><div class="HOEnZb"><div class="h5">> _______________________________________________<br>
> Molpro-user mailing list<br>
> <a href="mailto:Molpro-user@molpro.net">Molpro-user@molpro.net</a><br>
> <a href="http://www.molpro.net/mailman/listinfo/molpro-user" target="_blank">http://www.molpro.net/mailman/listinfo/molpro-user</a><br>
<br>
</div></div></blockquote></div><br>