Re: [AMBER] Error during production run - AMBER20

From: David A Case <david.case.rutgers.edu>
Date: Thu, 7 Apr 2022 06:53:56 -0400

On Thu, Apr 07, 2022, Sathyaseelan C wrote:
>
> I have installed AMBER20 with CUDA 9.0 and openmpi-3.0.0. I
>tried running the production run using pmemd.cuda. But it's showing the
>error which is given below. Any suggestion is highly appreciated.

Are you trying to use mpirun with pmemd.cuda? You would need
pmemd.cuda.MPI. But clearly, try a short serial CPU run, then a short
serial cuda run before jumping into parallel CUDA.
>
>cudaMemcpyToSymbol: SetSim copy to cSim failed invalid device symbol

What is your CUDA_VISIBLE_DEVICES set to?
Do the test cases work?

> Unit 9 Error on OPEN: pstep_1.rst

If this is a restart file, and a file with that name already exists, then
you need to have the -O flag set, or the program will refuse to overwrite
and existing file.

As you can see, with so little information, there are lots of things that
could be going wrong.

....good luck....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Apr 07 2022 - 04:00:02 PDT
Custom Search