[Wien] strange time using -it switch

[Yongsheng Zhang]

Dear all,

Currently, I use the latest Wien2k_08.1 version to compute a large
system on a computer with 4 CPUs. So I am trying the iterative
diagonalization (-it). The first iteration is full diagonalization, then
it is no problem that  the lapw1 in the my first iteration is about 4
hours. But in the second iteration, although it turns to -it, it doesn't
improve anything, and the lapw1 time is still about 4 hours. The
userguide says "-it" will save computer time factor of 2-5. Why it does
not have any improvement in my case?

My computational system has 8 k-points, and distribute the k-points on
each CPU properly.
Computer: Compaq
Fortran Compiler: f90
C compiler: cc

Here is the message in my case.dayfile

    cycle 1     (Tue Jan 15 22:46:59 CET 2008)  (99/99 to go)

>   lapw0 -p    (22:46:59) starting parallel lapw0 at Tue Jan 15
22:47:00 CET 20
08
--------
running lapw0 in single mode
353.262u 35.305s 6:29.31 99.8%  0+3385k 115+3729io 32pf+0w
>   lapw1  -p   (22:53:29) starting parallel lapw1 at Tue Jan 15
22:53:29 CET 20
08
->  starting parallel LAPW1 jobs at Tue Jan 15 22:53:29 CET 2008
running LAPW1 in parallel mode (using .machines)
4 number_of_parallel_jobs
     th12(2) 16143.807u 13.319s 4:30:03.26 99.7%        0+13876k
52+22077io 11pf
+0w
     th12(2) 16065.747u 15.477s 4:28:49.17 99.7%        0+13967k
14+19147io 6pf+
0w
     th12(2) 16078.786u 11.189s 4:28:45.91 99.7%        0+13881k
23+19198io 16pf
+0w
     th12(2) 16152.567u 9.806s 4:30:05.14 99.7% 0+13927k 0+19170io 0pf+0w
   Summary of lapw1para:
   th12  k=8     user=64440.9    wallclock=1076
28.743u 78.717s 4:30:24.20 0.6% 1+3k 481+4895io 39


    cycle 2     (Wed Jan 16 03:32:45 CET 2008)  (98/98 to go)

>   lapw0 -p    (03:32:45) starting parallel lapw0 at Wed Jan 16
03:32:46 CET 20
08
--------
running lapw0 in single mode
354.445u 35.066s 6:30.70 99.6%  0+2705k 8+3733io 3pf+0w
>   lapw1 -it -p        (03:39:16) starting parallel lapw1 at Wed Jan 16
03:39:1
6 CET 2008
->  starting parallel LAPW1 jobs at Wed Jan 16 03:39:17 CET 2008
running LAPW1 in parallel mode (using .machines)
4 number_of_parallel_jobs
     th12(2) 16151.818u 12.485s 4:30:06.36 99.7%        0+13877k
93+22080io 0pf+
0w
     th12(2) 16104.782u 11.022s 4:29:27.35 99.6%        0+13966k
33+19174io 0pf+
0w
     th12(2) 16141.763u 12.255s 4:29:55.10 99.7%        0+13881k
33+19207io 0pf+
0w
     th12(2) 16105.988u 11.590s 4:29:10.42 99.7%        0+13925k
36+19169io 0pf+
0w
   Summary of lapw1para:
   th12  k=8     user=64504.4    wallclock=1077
28.091u 80.072s 4:30:21.92 0.6% 1+3k 292+4897io 246pf+0w

Thanks
Zhang

-- 
---------------------------------------------------------------------
Address:  Fritz-Haber-Institut, Abt. Theorie
          Faradayweg 4-6 D-14195 Berlin (Germany)
Phone:    +49 30 8413 4818
Fax:      +49 30 8413 4701
Email:    zhang at fhi-berlin.mpg.de
---------------------------------------------------------------------
1-0.0735-11600-23.05