On Sat, Mar 29, 2008, Lachele Foley wrote:
>
> When I first wrote, I'd only gotten AmberTools to install. I still don't
> have the main amber10 package installed on the i686 systems (a newer
> gcc/gfortran should work, and I'll get busy installing that...). But, I
> got it all installed on the ia64. I saved the test results from all the
> different tests into separate files (attached tar.bz2) so you can see
> for yourself what's up. The file names should explain. There are more
> "formatting" (?) issues on the ia64 than on the i686 system. I don't know
> how important some of them are. I'm sure, for example, that the "%s" in
> the middle of the jaguar file header is insignificant (this happens both
> systems). But, I don't know if this difference (ia64 only) is ok:
Well, it looks like there are real problems with gfortran on an ia64 system!
There are lots of entries like this:
possible FAILURE: check hip.out.dif
/usr/local/programs/amber10/test/pbsa_radi
but there are no diffs listed. Is there in fact a non-zero length
"hip.out.dif" file? The script is not supposed to say "FAILURE" without
providing the diffs. Note that almost all of the "FAILURES" on QMMM codes are
of this sort (i.e. no diffs are provided in the TEST_FAILURES.diff file)
If you run one of these tests "by hand", is there a program crash, and errors,
etc?
Something else is troubling: You have lots of diffs that include this:
possible FAILURE: check mdout.dif
/usr/local/programs/amber10/test/ncsu/pmd
2c2
< Amber 10 SANDER 2008
---
> Amber 9 SANDER 2006
This means that the "save" file had the Amber 10 header, but that the file
you ran had an Amber 9 header. This in turn means that that somehow you
are not really running the current Amber10 codes. Can you double check your
environment variables and paths, etc.? I don't quite know how to parse this,
but it doesn't really make sense.
>
> possible FAILURE: check 1b2t.mol2.dif
> /usr/local/programs/amber10/test/antechamber/chemokine
> 26c26
> < 20 N 26.668 2.825 1.285 N.am 1 GLN 0.
> ---
> > 20 N 26.668 2.825 1.285 N.am* 1 GLN 0.
> ... (skip a few similar lines)
> 49c49
> < 43 ND1 31.505 5.900 2.809 N.pl3 2 HIS 0.
> ---
> > 43 ND1 31.505 5.900 2.809 N.pl* 2 HIS 0.
Confirmed, entered as bug 62, and I sent email to Junmei.
If you can't use ifort on your system, my guess is that g95 is more likely to
work, but I haven't tried that.
Sorry for the long email, but the things you report don't show up anywhere
else. And, for some reason, the TEST_FAILURES file is either reporting
FAILURES where there aren't any, or it is not collecting the failure
evidence the way it should.
...thanks!....dave
Received on Fri Apr 18 2008 - 21:15:21 PDT
