[AMBER] Partitioning issue of sander

From: 板界6柑哲彦 <itagaki08.gmail.com>
Date: Mon, 16 Apr 2012 11:00:14 +0900

Dear AMBER users and developers,

I would like to ask partitioning issue of sander. I noticed that similar

questions invoked in this ML forum, but could not find the solution.

So, please let me allow to invoke this issue again.


If you know the page providing the solution, please let me know just

the page in this ML.


I used sander to perform MD of protein using 8-core parallel computer
system, and detected the following error message;


----------------------------------------

 partition error in shake on processor 0
 this processor has atoms 1 through 9100
 atom 9098 is within this range
 atom 9101 is not within this range !

------------------------------------------


In the case of using 4 cores, sander works well, but for more than 8
cores, it does not, as mentioned (i.e., the above-mentioned error

message appears).

This system includes 16,799 water molecules. (Case 1)



Next, I added +3,989 water molecules to this model system by using

SolvateBox in LEaP, and noticed that sander worked well on a 16-core

parallel computer system, and that it did not on a 32-core parallel

system. (Case 2)


Then, I added +28,313 water molecules to this model system by

LEaP, and noticed that sander worked well even on a 32-core

parallel system. But it takes much more time than the Case 2.
(Case 3)

For this model system (Case 3), I removed 108 waters to build a
smaller calculation system, and tried the calculation again; as a result
of this modification, sander did not work on 32-core parallel
computer system (i.e., the above-mentioned error message appears).

Thus, the number of available cores looks very sensitive to sander
for this protein.

I would like to use 32-core parallel system even when using calculation
models 1 and/or 2 (Cases 1 and/or 2).

If you could let me know some instructions of this issue, I would much
appreciate it.

best regards,
Tetsuhiko
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun Apr 15 2012 - 19:30:03 PDT
Custom Search