24小时热门版块排行榜

>论坛更新日志 (3936)
>虫友互识 (613)
>文献求助 (488)
>导师招生 (366)
>考博 (204)
>休闲灌水 (190)
>论文投稿 (183)
>硕博家园 (152)
>基金申请 (117)
>招聘信息布告栏 (97)
>博后之家 (83)
>教师之家 (67)
>论文道贺祈福 (60)
>公派出国 (48)
>考研 (44)
>找工作 (41)

返回列表

当前主题已经存档。

imation

铁杆木虫 (正式写手)

应助: 0 (幼儿园)
金币: 7082.3
散金: 2
红花: 1
帖子: 829
在线: 345.4小时
虫号: 20276
注册: 2003-08-02
专业: 生物医用高分子材料

[交流] castep多核运算的问题

大家好，我有一台服务器，cpu为至强clovertown双路四核，每个核心频率是2ghz，内存667，4条，单条1g，操作系统装过2003中文版，sp1和sp2，也装过英文版sp1，2008英文版sp1
我尝试着运行tutorial里面的co分子adsorption在pb表面，一个核心运行很正常，1个小时就有结果，可是8个核心就总是失败，出现的问题也是千奇百怪：
1）进程castep，共八个，可是运行一会儿以后，自动减少到5个，有一次甚至减少到1个，内存无释放，有时任务失败，有时一直运算，但结果遥遥无期
2）cpu利用率在开始计算时为满负荷，经过1到5分钟，降到1~3%，内存没有释放，过10分钟左右，任务失败
3）尝试改变核心数目，发现1到3个都可以正常运行得到结果，超过3个就失败
4）还有n多错误，其中出现最多一个（每次都是在第一个点时出现），代码如下：

*Warning* max. SCF cycles performed but system has not reached the groundstate.

Current total energy, E    =  -5951.893478460    eV
Current free energy (E-TS)  =  -5951.990514506    eV
(energies not corrected for finite basis set)

NB est. 0K energy (E-0.5TS)    =  -5951.941996483    eV

  ****************************************************************************
  Warning: electronic minimisation did not converge when finding ground state.
  ****************************************************************************

Writing model to 1.check
Error in geom_get_forces - electronic_minimisation of current_cell failed
Error in geom_get_forces - electronic_minimisation of current_cell failed
Error in geom_get_forces - electronic_minimisation of current_cell failed
Error in geom_get_forces - electronic_minimisation of current_cell failed
[1] MPI Abort by user Aborting program !
[1] Aborting program!
[2] MPI Abort by user Aborting program !
[2] Aborting program!
[0] MPI Abort by user Aborting program !
[0] Aborting program!
forrtl: severe (47): write to READONLY file, unit 60, file D:\PROGRA~1\Accelrys\MATERI~1.1\Gateway\root_default\dsd\jobs\4GLA1\killfile
Image             PC       Routine          Line       Source
castepexe_mpi.exe  00AB6CC2  Unknown             Unknown  Unknown
castepexe_mpi.exe  00AB3F50  Unknown             Unknown  Unknown
castepexe_mpi.exe  00A42B9E  Unknown             Unknown  Unknown
castepexe_mpi.exe  00A427BB  Unknown             Unknown  Unknown
castepexe_mpi.exe  00A246D1  Unknown             Unknown  Unknown
castepexe_mpi.exe  009F8E7B  Unknown             Unknown  Unknown
castepexe_mpi.exe  009F8EAF  Unknown             Unknown  Unknown
castepexe_mpi.exe  004FB352  Unknown             Unknown  Unknown
castepexe_mpi.exe  004E5281  Unknown             Unknown  Unknown
castepexe_mpi.exe  004DC8D0  Unknown             Unknown  Unknown
castepexe_mpi.exe  00402353  Unknown             Unknown  Unknown
castepexe_mpi.exe  00ABE578  Unknown             Unknown  Unknown
castepexe_mpi.exe  00A904BB  Unknown             Unknown  Unknown
kernel32.dll    7C82F23B  Unknown             Unknown  Unknown

我尝试增加maximum iterations到1000，Max SCF cycles到5000，八核并行运算，算了2天半，正常，我stop运算，在输出中发现还是有些错误：
WARNING - user ionic constraints and symmetry specified
      - symmetry has precedence over constraints
      - may lead to a conflict?

HINT - if convergence fails try switching symmetry OFF

BFGS: Warning - trial step suggests complex energy landscape
               in which simple line minimization will fail.

            - This is usually an indication that the forces/streses
               are not accurate enough. Consider increasing the cutoff
               energy and/or the electronic convergence tolerance.

            - Proceeding with a bisection search to find root instead.

Warning: There are no empty bands for at least one kpoint and spin; this may
      slow the convergence and/or lead to an inaccurate groundstate.
      If this warning persists, you should consider increasing nextra_bands
      and/or reducing smearing_width in the param file.
      Recommend using nextra_bands of 14 to 29.

BFGS: Warning - Repeated consecutive reset of inverse Hessian
BFGS:          without satisfying convergence criteria which
BFGS:          looks like BFGS has run out of search directions.
BFGS: Warning - Lets try allowing some uphill steps and see if
BFGS:          we can get around this barrier.
BFGS: Warning - It is possible that the system may now converge to
BFGS:          a stationary point OTHER than the desired minimum.
BFGS: Hint - this may be an indication that either:
BFGS:          a) you are using a poor guess at geom_frequency_est
BFGS:          and/or geom_modulus_est, or
BFGS:          b) you are using unrealistic convergence criteria.
BFGS:          Suggest therefore that you consider changing them!

等等，就不一一列出了
我的问题是，为何单核1个小时就能算出来的project，多核要么不能运算，要么运算起来比单核还要慢得多？是不是多核并行处理的时候数据交换有问题啊？

谢谢，不知道描叙清楚了没有

回复此楼

» 猜你喜欢

拟解决的关键科学问题还要不要写已经有8人回复
存款400万可以在学校里躺平吗已经有28人回复
最失望的一年已经有11人回复
求推荐英文EI期刊已经有5人回复
请教限项目规定已经有4人回复
国自然申请面上模板最新2026版出了吗？已经有20人回复
26申博已经有3人回复
基金委咋了？2026年的指南还没有出来？已经有10人回复
基金申报已经有6人回复
疑惑？已经有5人回复

» 本主题相关商家推荐: (我也要在这里推广)

1楼 2007-12-07 18:46:41

已阅回复此楼关注TA 给TA发消息送TA红花 TA的回帖

carlon

金虫 (小有名气)

应助: 0 (幼儿园)
金币: 436.1
帖子: 155
在线: 62.4小时
虫号: 190669
注册: 2006-02-20
性别: GG
专业: 物理无机化学

★ ★
csfn(金币+2,VIP+0):感谢积极的交流 :-) 欢迎常来

多个cpu运行的话，容易出错。只要其中一个进程出错，计算就不会有结果
此时其他进程仍在运行，这可能就是为什么1个cpu1个小时算完，多cpu运行算很久没有结果的原因。
另外，在多个cpu运行过程中，上面还有很多warning信息
1。HINT - if convergence fails try switching symmetry OFF
这里可能是限制对称性，导致不好收敛，或者给原子设置了constrain，有时候收敛困难，可以在计算的时候去掉symmetry，试试看。
2。BFGS: Warning - trial step suggests complex energy landscape
in which simple line minimization will fail.
平面波cutoff设置可能过小。
3。empty band空带数不足，如果体系是金属性的话，不容易收敛。