[AMBER] Seg fault with pmemd.cuda.MPI using plumed

From: Jacob Remington <Jacob.Remington.uvm.edu>
Date: Sat, 11 Jul 2020 19:16:54 +0000

I am testing using amber20 to run umbrella sampling REMD with plumed (I am assuming this is a supported feature now with amber20?) and ran into errors[1] that I could reproduce in a much simpler normal MD run with pmemd.cuda.MPI, plumed=1, and a plumedfile that contained no applied biases. The same system (w and w/out biases on the cv using plumed) works great with sander.MPI, pmemd.MPI (with np = 2*ng), and pmemd.cuda, and only fails using pmemd.cuda.MPI. Without plumed (plumed=0) the same system also works fine with pmemd.cuda.MPI. I have two gpus on my workstation and am calling "mpirun -np 2 pmemd.cuda.MPI ...". Does anyone know what could be the issue?

Thank you for your time and effort,

Jacob M Remington

Postdoctoral Associate
University of Vermont





[1]
+++ Loading the PLUMED kernel runtime +++
+++ PLUMED_KERNEL="/usr/local/lib/libplumedKernel.so" +++
+++ Loading the PLUMED kernel runtime +++
+++ PLUMED_KERNEL="/usr/local/lib/libplumedKernel.so" +++
PLUMED: PLUMED is starting
PLUMED: Version: 2.6.0 (git: Unknown) compiled on Jul 3 2020 at 12:35:25
PLUMED: Please cite these papers when using PLUMED [1][2]
PLUMED: For further information see the PLUMED web page at http://www.plumed.org
PLUMED: Root: /usr/local/lib/plumed
PLUMED: For installed feature, see /usr/local/lib/plumed/src/config/config.txt
PLUMED: Molecular dynamics engine: amber
PLUMED: Precision of reals: 8
PLUMED: Running over 2 nodes
PLUMED: Number of threads: 1
PLUMED: Cache line size: 512
PLUMED: Number of atoms: 46354
... my cv information ...

PLUMED: Finished setup

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7f5c7333732a
#1 0x7f5c73336503
#2 0x7f5c71aa4fcf
#3 0x0
#0 0x7ffbc7c9332a
#1 0x7ffbc7c92503
#2 0x7ffbc6400fcf
#3 0x0
--------------------------------------------------------------------------
mpirun noticed that process rank 1 with PID 0 on node oula exited on signal 11 (Segmentation fault).
--------------------------------------------------------------------------


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber


Received on Sat Jul 11 2020 - 12:30:03 PDT
Custom Search