Intel E5520 chip / regtests

[Jörg Saßmannshausen]

Dear all,

I have recently aquired an Intel dual quadcore E5520 machine with 24 GB RAM 
and I am in the process of installing cp2k on it as I want to use the machine 
mainly for that purpose. I was wondering if somebody already has such a 
machine and happens to have some useful advice regarding compiler flags.
Right now I am using the following flags:
FCFLAGS  = $(DFLAGS) -I$(INTEL_INC) -I$(FFTW3_INCL) -O2 -xW -heap-arrays 
64 -funroll-loops -fpp -free
FCFLAGS2 = $(DFLAGS) -I$(INTEL_INC) -I$(FFTW3_INCL) -O1 -xW -heap-arrays 
64 -fpp -free

which are basically the ones supplied with the code. I only lowered the -O3 
to -O2 as I thought that makes more sense. 
Compiler is the 10.1.021 which I have used sucessfully in the past. I am using 
the Intel mkl 11.0.083 (which was part of the compiler suite we purchased). I 
could not use the Intel scalapack so I had to build my own scalapack/blacs 
library.

The regest run ok, however, I get some segfaults in the *emd* part of the 
tests. For example:

/usr/local/src/cp2k/rt/TEST-Linux-x86-64-intel-popt-2009-06-26T18:06:04+0100/QS/regtest-rtp/H2-emd-efield.inp.out
 ENERGY| Total FORCE_EVAL ( QS ) energy 
(a.u.):               -0.843928297900146
 MD_ENERGIES| Initialization proceeding

 ******************************** GO CP2K GO! 
**********************************
 INITIAL POTENTIAL ENERGY[hartree]     
=                     -0.843928297900E+00
 INITIAL KINETIC ENERGY[hartree]       =                      
0.142506690448E-02
 INITIAL TEMPERATURE[K]                =                                 
300.000
 INITIAL VOLUME[bohr^3]                =                      
0.145764026995E+04
 INITIAL CELL LNTHS[bohr]   =      0.1133836E+02   0.1133836E+02   
0.1133836E+02
 INITIAL CELL ANGLS[deg]    =      0.9000000E+02   0.9000000E+02   
0.9000000E+02
 ******************************** GO CP2K GO! 
**********************************
  Information at iteration step:           1
  Total electronic density 
(r-space):          -2.0000040568       -0.0000040568
  Total 
energy:                                                    -0.3030742537
  Energy difference to previous iteration step:                     
0.5408540442
  Convergence:                                                  
0.1000000000E+01
[taurus:02262] *** Process received signal ***
[taurus:02262] Signal: Segmentation fault (11)
[taurus:02262] Signal code: Address not mapped (1)
[taurus:02262] Failing at address: 0x311e254f68
[taurus:02262] [ 0] /lib/libpthread.so.0 [0x7f228e0a1a80]
[taurus:02262] [ 
1] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(rt_propagation_utils_mp_calc_c_mat_force_+0x153b) 
[0xaf25b1]
[taurus:02262] [ 
2] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(qs_force_mp_qs_forces_+0x5060) 
[0x8b6e4c]
[taurus:02262] [ 
3] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(force_env_methods_mp_force_env_calc_energy_force_+0x1ad3) 
[0x52dacd]
[taurus:02262] [ 
4] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(integrator_mp_nve_+0xb5c) 
[0x13a3d0a]
[taurus:02262] [ 
5] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(velocity_verlet_control_mp_velocity_verlet_+0x24b) 
[0xc1337b]
[taurus:02262] [ 
6] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(md_run_mp_qs_mol_dyn_+0x123d) 
[0x73fe95]
[taurus:02262] [ 
7] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(cp2k_runs_mp_cp2k_run_+0x12e8) 
[0x49991e]
[taurus:02262] [ 
8] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(cp2k_runs_mp_run_input_+0xe8) 
[0x4985d4]
[taurus:02262] [ 
9] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(MAIN__+0x1238) 
[0x4973dc]
[taurus:02262] 
[10] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt(main+0x2a) 
[0x496196]
[taurus:02262] [11] /lib/libc.so.6(__libc_start_main+0xe6) [0x7f228c61b1a6]
[taurus:02262] 
[12] /usr/local/src/cp2k/rt/cp2k/exe/Linux-x86-64-intel/cp2k.popt [0x4960a9]

Has anybody similar problems? I downloaded the latest version of cp2k 
yesterday.

All the best from Glasgow!

Jörg
-- 
*************************************************************
Jörg Saßmannshausen
Research Fellow
University of Strathclyde
Department of Pure and Applied Chemistry
295 Cathedral St.
Glasgow
G1 1XL

email: jorg.sassm... at strath.ac.uk
web: http://sassy.formativ.net

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html