amber-developers: known problems with the 20mar08 tarballs

From: David Case <case.scripps.edu>
Date: Fri, 21 Mar 2008 12:01:13 -0800

This is an addendum to the earlier email about test the 20mar08 stuff.

We primarily tested with g95/gcc3 on x86_64 architecture. There are some
known problems there that you don't need to re-report:

1. two failures in the antechamber/database tests (frcmod and prepi).

2. The LES_TEMP "2temp.out" test fails: the total energy printed on the first
step is (incorrectly) zero.

3. There are some round-off errors in some evb tests.

Also, we have fixed missing config files in pmemd.amba, and fixed the fact
that the AmberTools tests will fail on machines where g77 is not present.

It would be great is someone with access to gfortran could check if mopac
and resp can be built with it. Note that you probably need the equivalent
of the -fno-automatic -finit-local-zero flags that g77 knows about, in order
to get reliably correct results.

...thanks!...dave
Received on Fri Apr 18 2008 - 21:12:58 PDT
Custom Search