Re: [AMBER] amber Computational problem

From: David A Case via AMBER <amber.ambermd.org>
Date: Thu, 27 Oct 2022 08:54:31 -0400

On Thu, Oct 27, 2022, Ding Shuaiwei via AMBER wrote:
>
>您好:
> 有个问题想请教一下,我在超算集群环境中使用ambertools22的sander.MPI命令计算文件,输出结果中发现:
>如果用少量核数运行,比如 mpirun -np 32 sander.MPI -ng 8 -groupfile ./groupfile ,可以得到正常的计算结果
>但是如果增加运行的核数,比如 mpirun -np 64 sander.MPI -ng 8 -groupfile ./groupfile,输出的output文件,会说体系内原子的速度异常,数值太高,显示为一堆星号,算到最后值从一堆星号变为NaN
>使用的计算环境和文件都相同,唯一的区别是计算核数不同。
>请问一下造成这样问题的原因可能是什么呢?
>感谢!
>

It is likely that your system is either not minimized correctly, or that
there is some other problem. Try a short run, say nstlim=10, with ntpr=1.
That may provide more information.

....dac

_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Oct 27 2022 - 06:00:02 PDT
Custom Search