[交流] 【求助成功】段错误的解决方法是不是只有重新编译

Originally posted by hedaors at 2010-11-10 22:55:58:
Ba 的原子要比 Ca 的多很多，内存就可能不够。
这个在要求精度不变的情况下，只能牺牲速度，你用的是几个核心？你只用一个核心提交任务试试，应该没问题。

running on    1 nodes
distr:  one band on    1 nodes,    1 groups
vasp.4.6.28 25Jul05 complex
POSCAR found :  4 types and   80 ions
LDA part: xc-table for Pade appr. of Perdew
POSCAR, INCAR and KPOINTS ok, starting setup
WARNING: wrap around errors must be expected
[node11:02919] *** Process received signal ***
[node11:02919] Signal: Segmentation fault (11)
[node11:02919] Signal code: Address not mapped (1)
[node11:02919] Failing at address: 0x351761ab8
[node11:02919] [ 0] /lib64/tls/libpthread.so.0 [0x317210c430]
[node11:02919] [ 1] /home/software/openmpi-1.2.2-intel9/lib/libopen-pal.so.0 [0x2a966ea7f0]
[node11:02919] [ 2] /home/software/openmpi-1.2.2-intel9/lib/libopen-pal.so.0(_int_malloc+0x627) [0x2a966ea07f]
[node11:02919] [ 3] /home/software/openmpi-1.2.2-intel9/lib/libopen-pal.so.0(malloc+0x67) [0x2a966e93bf]
[node11:02919] [ 4] /home/bin/vasp.openmpi(for_allocate+0x89) [0x699c01]
[node11:02919] [ 5] /home/bin/vasp.openmpi(wave_mp_allocw_+0x2c7) [0x4a9977]
[node11:02919] [ 6] /home/bin/vasp.openmpi(MAIN__+0x1504d) [0x428b8d]
[node11:02919] [ 7] /home/bin/vasp.openmpi(main+0x32) [0x413b2a]
[node11:02919] [ 8] /lib64/tls/libc.so.6(__libc_start_main+0xdb) [0x317161c3fb]
[node11:02919] [ 9] /home/bin/vasp.openmpi [0x413a6a]
[node11:02919] *** End of error message ***
mpirun noticed that job rank 0 with PID 2919 on node node11 exited on signal 11 (Segmentation fault).

Originally posted by hakuna at 2010-10-27 16:29:54:
我16G内存也会出现这问题！
你大概是用intel的编译器干活的，当体系一大，刚进入主循环就会退出来，并提示上面的错误！前面提到降低截断能、减少K点主要是降低内存的消耗，如果是这样的的话，那便是内存溢出了，原因在于编译时INTEL默认将缓存写在堆栈里，堆栈小时便会出现上述错误，你可以尝试在编译VASP时，在FFLAGS里加上参数-heap-arrays  64试试看，如果还不行的话，就换一下编译器，PGI应该会好点。
祝你好运！别忘了报道一下你的结果啊

Originally posted by xianggui7895 at 2010-10-31 19:31:46:
你说的应该是正解。现在我找到一条折衷的解决方法，就是把PREC设为Medium之后就能算了，之前用的是PREC=Accurate……
[ Last edited by xian ...

Originally posted by valenhou001 at 2010-11-11 10:04:48:
是的。唯一的正确的解决方法就是重新编译

Originally posted by hakuna at 2010-10-27 16:29:54:
我16G内存也会出现这问题！
你大概是用intel的编译器干活的，当体系一大，刚进入主循环就会退出来，并提示上面的错误！前面提到降低截断能、减少K点主要是降低内存的消耗，如果是这样的的话，那便是内存溢出了，原因在于编译时INTEL默认将缓存写在堆栈里，堆栈小时便会出现上述错误，你可以尝试在编译VASP时，在FFLAGS里加上参数-heap-arrays  64试试看，如果还不行的话，就换一下编译器，PGI应该会好点。
祝你好运！别忘了报道一下你的结果啊