[Pw_forum] restart problem using Grid parallelization

[Xiaolong Zhang]

Dear all,

I've been doing electron-phonon-coupling calculation  on a Hexagonal
 system using the newly incorparated Grid parallelization method in
QE5.0.1,   the total q points in irr BZ in my case are 7, so naturally I
split the el-ph calculation into 7 individual job using the <<start_q,
last_q>> flags and submit to the cluster simultaneously.

Everything goes well except  when it comes to the  restart part(have to
restart due to the limited wall time on cluster).  The problem is when the
job restarted successfully from previous run for  q # 2 to # 7, the q#1
part has to be calculated in these q(in my first run, the job for q#1
hasn't finished), i.e., for these q(2-7), the job seems can't continue to
run separately as I expected.

Here is part of output for q#2 from restart run.
__________________________________________________________________________
__________________________________________________________________________
 Parallelization info
     --------------------
     sticks:   dense  smooth     PW     G-vecs:    dense   smooth      PW
     Min          82      82     24                27912    27912    4341
     Max          83      83     26                27959    27959    4371
     Sum         661     661    199               223537   223537   34849


     Check: negative/imaginary core charge=   -0.000001    0.000000
        2 /   7 q-points for this run, from  1 to  2:
       N       xq(1)         xq(2)         xq(3)
       1   0.000000000   0.000000000   0.000000000
       2   0.000000000   0.192450090   0.000000000
       3   0.000000000   0.384900179   0.000000000
       4   0.000000000  -0.577350269   0.000000000
       5   0.166666667   0.288675135   0.000000000
       6   0.166666667   0.481125224   0.000000000
       7   0.333333333   0.577350269   0.000000000


     Calculation of q =    0.0000000   0.0000000   0.0000000

     Modes are read from file
____________________________________________________________________--
______________________________________________________________________

I wonder if this is a bug or actually the code intend to behave like this,
any ideas will be greatly approciated, thanks in advanve.

all the best

Xiao-long Zhang
Institute of Physics, CAS
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20121220/aac1a53a/attachment.html>