Intel MKL now has a version that uses 64-bit integers (ilp64), as well as the AMD ACML.<br><br>I was trying to run a CISD calculation with more than 32 active orbitals, but it failed:<br><br>1PROGRAM * CI (Multireference internally contracted CI) Authors: H.-J. Werner, P.J. Knowles, 1987<br>
<br> TOO MANY ACTIVE ORBITALS: 36 THIS VERSION ALLOWS 32<br><br>I was running on an em64t platform with -i8 option. Is there any way to solve this problem?<br><br>Thanks,<br><br><div class="gmail_quote">On Feb 1, 2008 1:17 AM, Andy May <<a href="mailto:MayAJ1@cardiff.ac.uk">MayAJ1@cardiff.ac.uk</a>> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Daniel,<br><br>You are correct, the -i8 option does not appear to be respected by<br>configure in this case. I guess the option only works on certain<br>
systems, I will take a look at this, but I do not think this is easy to<br>fix. The good new is that the problem does not exist in the developer<br>build so any future release will fully respect this. The bad news is so<br>
much has changed since it is impossible to backport this directly.<br><br>Note, if this had worked you would not be able to use your MKL BLAS<br>library since it contains 4 byte integers.<br><br>Best wishes,<br><font color="#888888"><br>
Andy<br></font><div><div></div><div class="Wj3C7c"><br>Daniel Brue wrote:<br>> Dear Molpro users,<br>> I am trying to compile molpro 2006.1 from source using the Intel Fortran<br>> compiler v10 on a Dell precision with dual Xeon 32 bit processors and<br>
> using Intel MKL libraries v9. I am trying to compile molpro to run with<br>> 32 active orbitals for mcscf calculations.<br>><br>> Other instances of this question arising in the mailing list have said<br>> that this was not possible for 32 bit processors, however, we were able<br>
> to compile molpro 2002.6 to run with 32 active orbitals and used this<br>> successfully for some years.<br>><br>> Is it possible to get 32 active orbitals with the 2006 version of<br>> molpro? And if not, what is the limitation? If it can be done, how does<br>
> one compile it to make this work?<br>><br>> Running the configure script with the -i8 option appears to ignore the<br>> option and the -i8 flag is not used in compiling.<br>><br>> Thanks for any help that anyone can offer.<br>
><br>> Sincerely,<br>> Daniel<br>><br>><br>> Daniel Brue<br>> Homer L. Dodge Dept. of Physics and Astronomy<br>> University of Oklahoma<br>> 440 W. Brooks St.<br>> Norman, OK 73019<br>> <a href="http://nhn.ou.edu/%7Ebrue" target="_blank">http://nhn.ou.edu/~brue</a><br>
> <a href="mailto:brue@nhn.ou.edu">brue@nhn.ou.edu</a><br><br></div></div></blockquote></div><br><br clear="all"><br>-- <br>Shenggang Li