AMBER Archive (2008)

Subject: RE: AMBER: Test failure with Amber 10

From: Kevin Parkes (Kevin.Parkes_at_medivir.com)
Date: Wed Nov 26 2008 - 08:36:36 CST


> On Wed, Nov 26, 2008, Kevin Parkes wrote:
> >
> > I'm trying to build Amber 10 on x86_65 linux with gcc /
> gfortran 4.2.3.
> > The process proceeds smoothly but testing generates two
> related erors:
> >
> > possible FAILURE: check mdout.tip5p.dif
> > /usr/local/medivir/amber/amber10/test/tip5p
> > 180,181d179
> > < EKCMT = 0.2952 VIRIAL = 9.6699 VOLUME =
> > 0.0001
> > < Density =
> > 0.
> > ---------------------------------------
> > possible FAILURE: check amoeba_wat2.out.dif
> > /usr/local/medivir/amber/amber10/test/amoeba_wat2
> > 206a207
> > > EKCMT = 0. VIRIAL = 0. VOLUME =
> > 0.0002
> > 208a210
> > > Density =
> > 0.
>
> Both differences are innocuous, I think. They have to do
> with roundoff errors leading to a different printout of RMS
> fluctuations of a very short run. We should probably make
> the tests work better, but I don't think these are cause for concern.
>
> ...dac
>

Digging a little more I agree that the differences appear to be
innocuous. Manual comparison of my generated mdout.tip5p and the
mdout.tip5p.save files shows that there is perfect agreement of the
results in the 10 steps of the simulation, the averages, and RMS. The
differences are due to the fact that the reference output is from a job
run with the PMEMD implementation of Sander (stated on line 6 of
mdout.tip5p.save) which outputs some additional information including
the fields highlighted in the test difference.

        Kevin Parkes
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo_at_scripps.edu