[molpro-user] MPPX version for SGI Altix

[Sarah Wilsey]

Hi,

I have been trying to get the MPPX version of Molpro running on our  
SGI Altix 4700. I have compiled it using the same options as I used  
for the MPP version except using -mppx instead of -mpp. The code  
compiles without any problems but when I run the job below, the  
optimization works fine but the frequencies produced are completely  
wrong. In addition to this, 2 and 4 processor jobs hang at the end  
when they try to release the memory with the error message:

1:kr_malloc: kr_free: block not found in used list
  2305843014037012096(0x200000011f7dfe80)
: 528350848
2:kr_malloc: kr_free: block not found in used list
  2305843018197761664(0x20000002177dfe80)
: 394133120


Input file:
***, SiH2Si Z-Matrix
memory,32,m
gthresh,energy=1.d-10,thrgrad=1.d-10
basis=vdz
   r1= 2.2145 ang
   a1= 52.3 degree
   r3= 1.2504 ang

   geometry={Si1
   Si2,Si1,r1
   Q1,Si1,r1,Si2,60
   Q2,Si1,0.5*r1,Si2,0,Q1,a1
   H1,Q2,r3,Si1,90,Q1,a1
   H2,Q2,r3,Si1,90,Q1,-a1}

mass,Si=27.976927,H=1.007825
mass,print
hf
ccsd(t)
optg,gradient=1.d-4
frequencies

Output file:
Frequencies dumped to record   5400.2

  Gradient norm at reference geometry: 0.40953D+05

  ?WARNING: GRADIENT NORM NOT ZERO!

  Normal Modes

                               1 A1        2 B1        3 B2        4 A1
  Wavenumbers [cm-1]     10216314.71 12986883.78 23082412.29 24778924.17
  Intensities [km/mol]          0.00        0.00        0.00        0.00
  Intensities [relative]        0.00        0.00        0.00        0.00
            SI1X1            0.00000     0.00000    -0.01997     0.00000
            SI1Y1           -0.08781    -0.02473     0.00000    -0.01362
            SI1Z1           -0.01863    -0.01671     0.00000    -0.00106
            SI2X2            0.00000     0.00000    -0.01997     0.00000
            SI2Y2            0.08781    -0.02473     0.00000     0.01362
            SI2Z2           -0.01863     0.01671     0.00000    -0.00106
            H1X3            -0.06989     0.00000     0.55439     0.70005
            H1Y3             0.00000     0.68657     0.00000     0.00000
            H1Z3             0.51726     0.00000    -0.42154     0.02934
            H2X4             0.06989     0.00000     0.55439    -0.70005
            H2Y4             0.00000     0.68657     0.00000     0.00000
            H2Z4             0.51726     0.00000     0.42154     0.02934


The code is patched to patch level 149, compiled with intel/cc/ 
0.1.044 and intel/fc/9.1039, and linked to cmkl-8.1/ilp64 and ga-4-0-1.

I would be grateful if anyone could let me know if they have got this  
to work on a similar architecture and if there are any other options  
I could use when compiling the code which might help,

Many thanks,

Sarah Wilsey


------------------------------------------------------------------------ 
--------------
Dr Sarah Wilsey
EPSRC National Service for Computational Chemistry Software
Department of Chemistry
Imperial College
London SW7 2AZ
Phone: 020 7594 1220      Fax: 020 7594 5804
URL: http://www.nsccs.ac.uk