Re: Sander: crash if nmropt=1. HELP!!!!!!!

From: David Konerding <dek_at_konerding.com>
Date: Tue 23 May 2000 10:33:20 -0700

Enrico Morelli writes:
>Dear,
>
>We had properly compile amber 6 under RH 6.1 using pgroup compiler ver
>3.1.
>
>We tried to repeat, with sander 6, an energy minimization using NOE
>constraints already done with sander 5 for a protein of 73 residues.

This error was reported:

NMR optimizations using sander compiled with the Portland Group
compiler "pgf77" fail with a floating point exception.

I have isolated the problem to a single file: nmrnrg.f. If nmrnrg.f
is compiled with optimization (it normally is) then sander will crash
when using NMR optimization.

If you compile it without optimization, it does not crash.

Therefore, if you are using PG compiler to compile sander, and you need
to work around the error, edit src/sander/Makefile, and right after the line:


#do not optimize nmrcal, due to bug on mips/SI compilers
nmrcal.o: nmrcal.f
                ../Compile L0 -P -D$(PREC) nmrcal.f


add the following:


#do not optimize nmrnrg, due to bug on Portland Group compiler
nmrnrg.o: nmrnrg.f
                ../Compile L0 -P -D$(PREC) nmrnrg.f


Unfortunately, I don't know the actual cause, but I don't think it's a bug in AMBER, since this
code apparently works fine with other compilers.

Dave
Received on Tue May 23 2000 - 10:33:20 PDT
Custom Search