AMBER Archive (2009)

Subject: Re: [AMBER] array out of bound in Amber10 tests

From: Shan-ho Tsai (
Date: Wed Feb 11 2009 - 09:38:23 CST

> These are probably not errors at all, but just reflect the fact that the
> bounds-checking code doesn't know how to deal with a older (but not
> incorrect) fortran syntax, analogous to the change discussed earlier.

Thank you very much for your response. Perhaps it is the case
that the compilers are able to assign the arrays properly, but
the bounds-checking code is not able to understand the older
syntax. It is true that without adding the bounds-checking code
the "cd 4096wat && ./Run.pure_wat_nmr_temp_reg" test passed
without any change to the nmr.f code. But it is a bit disturbing
to me that the bounds-checking code of both the Portland Group
pgf90 and the IBM xlf90 failed to deal with the syntax. And I
don't feel completely confident that the arrays are being
created correctly with these compilers. Anyhow, I can try to
make some simple test codes with the syntax and see what happens.

> The real question is this: if you turn off bounds checking, which test
> cases still fail?

If I turn off bounds checking, on the Linux cluster (using pgf90),
all serial and parallel Amber10 tests passed. However, on the AIX
cluster if I turn off bounds checking all serial tests pass, but
rem_hybrid runs until completion, and generates incorrect results.
The potential energy turned out to be around -1450 instead of -615.

Any suggestions how I can troubleshoot what is going wrong in the
rem_hybrid test?

Thank you so much!

AMBER mailing list