Re: [AMBER] Random segfaults (invalid memory reference) in sander

From: Charo del Genio <>
Date: Fri, 3 May 2019 15:08:31 +0100

On 03/05/2019 13:29, Daniel Roe wrote:
> Hmm - I think both of you are using OpenMPI. Do you have another MPI
> library (mpich, mvapich, intel mpi) that you can test with?
> -Dan
Hi Dan,
        I thought exactly the same thing, and in fact I've been trying. I have two other machines, where the code runs without problems (so far) using OpenMPI version 2.0.2. I've had it installed now on the
cluster as well, and I had no crashes for the last 16 hours. It may be still too early to say, but it looks like the specific version of OpenMPI may have something to do with it. Should it keep
failing, I'll recompile the code with -debug and see what that leads to. Either way, I will write again as soon as more information is available.



Dr. Charo I. del Genio
Senior Lecturer in Statistical Physics
Applied Mathematics Research Centre (AMRC)
Design Hub
Coventry University Technology Park
Coventry CV1 5FB
AMBER mailing list
Received on Fri May 03 2019 - 07:30:02 PDT
Custom Search