CCL: Pd-ethylene cluster with G98



 Hi netters,
 
I am comparing the performance of HF, DFT/B3LYP anda MP2 with Gaussian98 in a simple calculation of a cluster formed by two Pd atoms and ethylene. To my surprise, b3lyp is
 much slower than HF or MP2; cpu times are
 hf: 	3 min. 21 s.
 b3lyp:	1 h. 5 min.
 mp2:	8 min. 28 s.
 P-IV running LINUX
 this is my input file:
 ------------------------------------------------------------------------------------
 %chk=pd2etepa.chk
 # hf/lanl2mb opt(z-matrix,maxcyc=60) scf(qc,maxcyc=500)
 Title Card Required
 0  1
  Pd
  Pd                 1        2.750872
  C                  1              B1    2              A1
 
C 3 B2 1 A2 2 D1 H 4 B3 3 A3 1 D2 H 3 B4 4 A4 5 D3 H 4 B5 3 A5 6 D4 H 3 B6 4 A6 7 D5
    B1             2.408742
    B2             1.354929
    B3             1.070182
    B4             1.070143
    B5             1.070035
    B6             1.069670
    A1            76.672064
    A2           107.368541
    A3           120.230124
    A4           119.863439
    A5           119.916731
    A6           120.275618
    D1             1.099106
    D2            90.171155
    D3             0.053786
    D4          -179.938316
    D5             0.015838
 -------------------------------------------------------------------------------------------
 Is this normal?
 Can dft performance be improved?
                   \|/
               (   { *at * }   { *at * }   )
 -------oOo-(_)-oOo------------------------------------------------------
 Jeszs Rodrmguez-Otero
 Departamento de Qummica Fmsica, Facultade de Qummica
 Universidade de Santiago
 15782 Santiago de Compostela, (Spain)
 Phone: (+34)-981563100 ext 14436, Fax: (+34)-981595012
 qftjesus:at:usc.es
 http://web.usc.es/~qftjesus/index.html