On Thu, Oct 17, 2019, Tru Huynh wrote:
Just to add some trivial details about Amber's test procedure:
>Finished test suite for AmberTools at Tue Oct 15 22:13:35 CEST 2019.
>
> 2436 file comparisons passed
> 1 file comparisons failed ( 1 ignored)
> 5 tests experienced errors
>possible FAILURE: (ignored) check FINAL_RESULTS_MMPBSA_1.dat.dif
It was this FAILURE that led to the "ignored" statement in the summary.
We understand what is going on here, and it is correct to ignore this on
OSX. Of course, this is the lazy way out, and we should create a better
test case.
>the ignored was caused by:
>==============================================================
>cd ccif && ./Run.ccif
> ./Run.ccif: Program error
>make[3]: [test] Error 1 (ignored)
No: this "ignored" message comes from the Makefile, which is instructed
to soldier on (hence running all the tests) even if some of them fail.
I understand that this is confusing, since there are two ways the word
"ignored" can appear in the log file. Again, we just don't have enough
developers to squash all such problems.
>In all cases I get a FPE error for:
>cd dhfr && ./Run.dhfr.min_prcg
>Program received signal SIGFPE: Floating-point exception - erroneous arithmetic operation.
This is of course something that needs to be looked into; but the upside
is that it seems specific to using the PRCG minimization option in
sander, which very few people ever attempt.
Thanks again for doing all this work and providing us good feedback.
...dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Oct 18 2019 - 08:30:02 PDT
