Hi,
On Fri, 31 Mar 2006, David A. Case wrote:
> On Fri, Mar 31, 2006, Scott Brozell wrote:
> >
> > exe/README does not contain all the executables.
>
> I don't think this file shows up in the release.
ok, but it did show up in the snapshot.
> > Several tests are hanging.
> > The fix is to turn down the optimization of src/lapack/dlamch.f
> > I changed line 42 of src/lapack/Makefile to
> > $(FC) -c -o $. _$<
>
> Sounds like a better idea is to remove the "-O" from the fflags variable
> for pgf90 in configure. Does pgf90 have a "-O0" option?
Yeah, I should have been clearer.
I recommend
fflags="-O1 bla_bla"
where bla_bla is hardware dependent as you will see in configure.
Technically, I tested "-O1 -tp p7" which fixed the problem.
> It is well known that dlamch.f should not be optimized, and that was the
> *intent* of the special rule in the Makefile.
Since I don't recall a compiler that does not support -O0,
I suggest that the dlamch.f rule hard code -O0 in Amber 10.
> > ./configure -bintraj pgf90
> > pgf90 6.0-8 32-bit target on x86 Linux
> > Linux piv-login1 2.4.24-pre1 #6 SMP Wed May 19 09:47:10 EDT 2004 i686
i686 i386 GNU/Linux
> > GNU gdb Red Hat Linux (6.3.0.0-1.62rh)
> >
> >
> >
> > 2. cd tgtmd/change_target; ./Run.tgtmd
> > 3. cd tgtmd/change_target.rms; ./Run.tgtmd
> > 4. cd tgtmd/conserve_ene; ./Run.tgtmd
> > 5. cd tgtmd/minimize; ./Run.tgtmin
> > 6. cd tgtmd/PME; ./Run.tgtPME
> > 7. cd trajene; ./Run.trajene
> > 9. cd neb_gb; ./Run.neb.sa
> >
>
> Am I right in reading that the above are the remaining failures (mostly
> "hangs")?
I think these were all caused by dlamch.f.
And tomw I'll report whether the fix corrected them.
Scott
Received on Wed Apr 05 2006 - 23:49:31 PDT
