Thank you very much Andreas and Matthias for all the suggestions. At the moment however I'd like to stick as long as possible with the IBM compiler version installed on our machine (10.1), since I suppose the machine native compiler should give the best performance, in principle. Moreover, every time I have asked to the consultants in my HPC center about upgrading the compiler or installing a recent GCC, I have received no answer. Perhaps changing compiler looks to be an extreme solution and they're unwilling to do it... The current GCC version on our SP6 machine is 4.2.0, for which IBM binaries are available. Is it too obsolete for compiling ALPS 2 ? Is it possible to turn off a particular test, I mean the one in alea for which no workaround exist ? Just to see if there are any more issues ?
Best
Jessica
On Mer, 3 Novembre 2010 2:08 pm, Andreas Läuchli wrote:
Hi Jessica,
on the ibm machines it seems that HZ is a predefined macro, which renders the corresponding line in the looper code invalid.
perhaps one has to include an ibm workaround in the relevant looper files which looks roughly like the line below (I'm not sure about the capitalization though)
#ifdef HZ #undefine HZ #endif
Andreas
On Nov 3, 2010, at 6:02 AM, Matthias Troyer wrote:
Hi Jessica,
The best might really be if you upgrade to the latest version of the IBM compiler or use the GNU compiler instead.
For your comments, the first is a compiler issue which you will only solve by using a standard compliant compiler. There is no workaround for that compiler bug.
- The second error (related to applications) is:
"/gpfs/scratch/userexternal/jalfonsi/alps-2.0.0b4-r4959-src-with- boost/alps/applications/qmc/looper/looper/model_parameter.h", line 49.10: 1540-0063 (S) The text "100" is unexpected.
Here I do not see the text 100 in the current version. Can you please send the file? I would generally recommend that you upgrade to the latest nightly snapshot instead of the b4 release.
Matthias
andreas laeuchli /------------------------------------------------------------\ |vox: ++49 (0)351 871 11 15 | MPI PKS | |fax: ++49 (0)351 871 11 99 | Office 1 A 15 | |email: aml@pks.mpg.de | Noethnitzerstr. 38 | |http://www.pks.mpg.de/~aml | D-01187 Dresden, Germany | ------------------------------------------------------------/
-- Dr.ssa Jessica Alfonsi Dipartimento di Scienze Chimiche Universit� di Padova via Marzolo 1 35100 Padova (Italy) +39 0498275726 e-mail jessica.alfonsi@unipd.it