My Community

Hello,

I compiled vasp 4.6.27 mpi version with FFTW 3.0.1, ifce9.0.26 on Fedra Core 2 on AMD 64.
I could compile successfully. But It doesn't work.
Serial version works well in same environment with much better speed.
Mpi version with Juergen Furthmueller fft also works well (but it looks slow).

My FFTW compiling environment is,
- gcc 3.3.5
- g++ 3.3.5
without -m32 option. I tried both with '--enable-sse2' option and without this option.

MPICH1.2.7 is used to compile vasp. MPICH itself is compiled by gcc and ifce9.0.26.

When I run it, the STDOUT is like,

-----
running on 2 nodes
distr: one band on 1 nodes, 2 groups
vasp.4.6.27 26Jun05 complex
POSCAR found : 2 types and 4 ions
LDA part: xc-table for Ceperly-Alder, Vosko type interpolation para-ferro
POSCAR, INCAR and KPOINTS ok, starting setup
WARNING: wrap around errors must be expected
FFT: planning ... 1
p1_6172: p4_error: interrupt SIGSEGV: 11
rm_l_1_6174: (3.636719) net_send: could not write to fd=5, errno = 32
p1_6172: (3.636719) net_send: could not write to fd=5, errno = 32
p0_21491: p4_error: interrupt SIGx: 13
p0_21491: (5.742188) net_send: could not write to fd=5, errno = 32
-----

Would someone give me a good advise ?

Thank you,
<span class='smallblacktext'>[ Edited ]</span>

My vasp (series version 4.6 ) was working fine by using the fft included in vasp itself. Now I complied it again using fftw.3.0.1, it fails to work any more. The error message is as follows:

vasp.4.6.26 15Dec04 complex
POSCAR found : 2 types and 16 ions
LDA part: xc-table for Ceperly-Alder, standard interpolation
POSCAR, INCAR and KPOINTS ok, starting setup
FFT: planning ... 1
Segmentation fault


Could anyone help me out why this segmentation fault is?

I had the same problem, it was caused by fftw plans needing 64 bits storage on 64 bit systems. You could try my unofficial patch at

http://cms.mpi.univie.ac.at/vasp-forum/ ... .php?3.741

Note that the forum messed up the whitespace in the diffs somewhat.

Thank you. It works !