Hi Prof Renato,
I have also met the memory problem before. It could be caused by the HPC scheduler. In short, the GPU is not available although the job got scheduled. Have you tried to re-submit the job?
Best,
Lin
On Mar 29, 2021, at 8:47 AM, Renato Araujo <renatoacufpa.gmail.com<mailto:renatoacufpa.gmail.com>> wrote:
Dear Prof David
I’ve tried your suggestions and I’m still getting the same error
cudaMemcpy GpuBuffer :: Download failed an illegal memory access was
encountered
I did the balancing process in CPU.
200ps of heating, 300ps of density and 500ps of balance.
In the production stage I started to do with the GPU, but the error
happened again;
cudaMemcpy GpuBuffer :: Download failed an illegal memory access was
encountered
The error happens in the first 10ns of simulation.
Any other suggestions to resolve this error?
Note: This happens with only one complex (liganate-protein) in three other
complexes, this does not occur. The same protocol is used for everyone.
Em qua., 24 de mar. de 2021 às 12:07, David A Case <david.case.rutgers.edu<mailto:david.case.rutgers.edu>>
escreveu:
On Wed, Mar 24, 2021, Renato Araujo wrote:
Beyond that, one would need more information to help with debugging. Is
this happening in the middle of a long simulation, or right at the
beginning?
It happens at the beginning of the simulation, between 10 and 20ns.
That helps some, but look at the error: does it happen at very early step,
i.e. within the first 10 steps? You may need to run a short trajectory
with ntpr=1 to get this information.
Also, be sure that you have applied all updates: in particular, not
applying
update.8 may lead to instabilities if you are using the ff19SB force field.
...good luck....dac
_______________________________________________
AMBER mailing list
AMBER.ambermd.org<mailto:
AMBER.ambermd.org>
https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!HXCxUKc!m5c-Xq4J99U9PTR8YYLHpL8y0NTSP7Z6_K9lqaE0-4jnigRb4hg9W4mT73RWqt94DoxS9Lro$
--
Prof Dr Renato Costa
Instituto Federal do Pará - IFPA
Grupo de Modelagem Molecular - UFPA
Tel.+55 91 985484622
_______________________________________________
AMBER mailing list
AMBER.ambermd.org<mailto:
AMBER.ambermd.org>
https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!HXCxUKc!m5c-Xq4J99U9PTR8YYLHpL8y0NTSP7Z6_K9lqaE0-4jnigRb4hg9W4mT73RWqt94DoxS9Lro$
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Mar 29 2021 - 11:00:03 PDT
