I don't have cuda+amber experience, but I would make sure you have the
highest cuda version that your amber version supports. In general it
sounds like it could be a cuda bug or version mismatch, based on
'losing' a fixed amount of memory per cycle.
Bill
On 11/7/18 7:21 PM, M RCC wrote:
> Yes. I have three complexes (same protein with different Ligands). All
> three simulations were stopped exactly at 95 ns. Could not able to restart
> from the last .rst file also. Restarting also gives the same error.
>
> On Wed, 7 Nov 2018 at 3:02 PM, M RCC <mkr3j2c1.gmail.com> wrote:
>
>> Cuda Version is 7.5
>>
>> On Mon, Nov 5, 2018 at 12:40 PM M RCC <mkr3j2c1.gmail.com> wrote:
>>
>>> Dear Amber Users,
>>> My aMD calculation stopped after 95ns, with the following error
>>>
>>> "*cudaMemcpy GpuBuffer::Download failed an illegal memory access was *
>>> *encountered*"
>>>
>>> My simulation system contains protein 300aa + Inhibitor + 4 zinc atoms.
>>> I used cationic dummy atom approach to fix the zinc atoms.
>>>
>>> while setting the threshold value for aMD, I have included all atoms
>>> including zinc atoms.
>>> for setting the dihedral threshold also I have counted zinc atoms as
>>> residues since the dummy atom approach used.
>>> Simulation stopped with this error. But there is no error description was
>>> found in the log (.out) file.
>>>
>>> Is the error is due to the zinc atoms?
>>>
>>> How do I rectify this error?
>>>
>>>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Nov 07 2018 - 20:30:02 PST
