各位老师好,我是用超算平台Gromacs2020.6版本跑的模拟。由于GPU资源紧张,不确定能不能排上,也不确定哪个速度快,我就先用intel-MPI -np 40 纯CPU跑着,然后一边排GPU队列,一边不定时把前面产生的过程的文件(trr cpt log等)都复制到 GPU排队的任务目录下,排上后GPU -ntomp 30继续跑,这个是没问题的,现在还在正常跑着。
我看CPU40跑得就比GPU这边慢20%左右,我就想看一下80核是不是比GPU跑更快,就把GPU跑的文件再不定时复制到CPU 80核排队的目录下,排上后运行就出现了如下报错和提示,不知道是什么原因,和MPI核数有关吗?麻烦各位老师帮忙看一下。
运行命令:gmx_mpi mdrun -s md.tpr -cpi md.cpt -v -deffnm md -dlb
yesstarting mdrun 'Protein in water'
500000000 steps, 1000000.0 ps (continuing from step 174963600, 349927.2 ps).
-------------------------------------------------------
核心报错:Fatal error:
24 particles communicated to PME rank 13 are more than 2/3 times the cut-off
out of the domain decomposition cell of their charge group in dimension y.
This usually means that your system is not well equilibrated.
其他提示:
Compiled SIMD: AVX2_256, but for this host/run AVX_512 might be better (see
log).
Using 80 MPI processes
Non-default thread affinity set, disabling internal thread affinity
Using 1 OpenMP thread per MPI process