10 QA tests failed using downloaded binaries

Just Got Here

5:00:42 AM PST - Thu, Dec 8th 2011
Hi there, I just installed the nwchem-6.0-binary-redhat-5-5-gcc-4-1-2 binaries on two different machines with different operating systems. On both machines, the same 10 tests failed while all others passed. The QA tests which failed were: pyqa oh2 (this should fail, so not really a failure). dft_cr2 tce_cr_eom_t_ozone c60_ecp xe_zora_sf i2_zora_so o2_zora_so lys_qmmm ethane_qmmm qmmm_opt0 I am completely new to NWCHEM and would greatly appreciate some advice! - Why did these tests fail? - How can I remedy the problem so that all tests (bar the one which should fail) pass ? The first machine is running SUSE Linux Release 11 (Kernel 2.6.32.29-0.3-default) with Intel Xeon(R) processors. The second machine is running CentOS release 4.8 (Kernel 2.6.9-89.0.16.ELsmp) on an x86_64 platform. Please find the doqmtests.log file and the failed tests output files here: http://dl.dropbox.com/u/30958682/failedJobsOutput.tgz Many thanks in advance for your help, Mike

Forum Vet

3:44:37 PM PST - Thu, Dec 22nd 2011
Hi Mike, Looks like our test criteria are a little too stringent. Looking at the final results of the output files and comparing them to those in the test suite, everything seems fine. With the exception of: pyga: This require Python to be compiled into NWChem tce_cr_eom_t_ozone: You may want to run this test by explicitly running this on 4 processors. Bert Quote:Mwykes Dec 8th 12:00 pm Hi there, I just installed the nwchem-6.0-binary-redhat-5-5-gcc-4-1-2 binaries on two different machines with different operating systems. On both machines, the same 10 tests failed while all others passed. The QA tests which failed were: pyqa oh2 (this should fail, so not really a failure). dft_cr2 tce_cr_eom_t_ozone c60_ecp xe_zora_sf i2_zora_so o2_zora_so lys_qmmm ethane_qmmm qmmm_opt0 I am completely new to NWCHEM and would greatly appreciate some advice! - Why did these tests fail? - How can I remedy the problem so that all tests (bar the one which should fail) pass ? The first machine is running SUSE Linux Release 11 (Kernel 2.6.32.29-0.3-default) with Intel Xeon(R) processors. The second machine is running CentOS release 4.8 (Kernel 2.6.9-89.0.16.ELsmp) on an x86_64 platform. Please find the doqmtests.log file and the failed tests output files here: http://dl.dropbox.com/u/30958682/failedJobsOutput.tgz Many thanks in advance for your help, Mike

Guest -

3:09:02 PM PST - Mon, Jan 30th 2012
ZORA tests
It seems though, that ZORA implementation was broken in the version 6.0. So the _zora_ tests should have failed. I've tried some other ZORA computations with version 6.0, and they gave me strange results, few tens kcal/mol off. DK was OK, for the same test molecules. -- With best regards, Grigory Shamov

Forum Regular

10:40:33 AM PST - Fri, Feb 3rd 2012
Hi Grigory, Mike The ZORA implementation is not broken. Please try the 6.1 version. A number of new enhancements have been implemented in the current release. Small differences in the final ZORA energies are fine. For ZORA calculations I recommend using the following: set dft:tol_rho 1d-30 set int:cando_txs f set dft:job_grid_acc 1d-20 relativistic zora on zora:cutoff 1d-30 end Best, -Niri niri.govind@pnnl.gov

Forum >> NWChem's corner >> Compiling NWChem