about ntasks=6 in CAM4.0

[huangdzh@foxmail_com]

Dear eaton,
These days I was confused about the option -ntasks, I remenber that last month it can be -ntasks 6, but now it do not work with -natasks 6, I must set it -ntasks 4 or 2. Is there anything wrong with my mpi?
What's more, my conputer is a HP PC, the same problem with a linux cluster.
the message from the output said:
...
Domain Information

Horizontal domain: nx = 144
ny = 96
No. of categories: nc = 1
No. of ice layers: ni = 4
No. of snow layers:ns = 1
Processors: total = 1
Processor shape: square-pop
Distribution type: cartesian
Distribution weight: latitude
max_blocks = 1
Number of ghost cells: 1

CalcWorkPerBlock: Total blocks: 6 Ice blocks: 6 IceFree blocks: 0 Land blocks: 0
Processors (X x Y) = 1 x 1
Active processors: 1
(shr_sys_abort) ERROR: ice: no. blocks exceed max: increase max to 6
(shr_sys_abort) WARNING: calling shr_mpi_abort() and stopping
p0_28551: p4_error: : -1
p0_28545: p4_error: : -1
p0_28546: p4_error: : -1
p0_28549: p4_error: : -1
p0_28544: p4_error: : -1
p0_28548: p4_error: : -1
p0_28547: p4_error: : -1
p0_31767: p4_error: : -1
p0_31770: p4_error: : -1
p0_31766: p4_error: : -1
p0_31768: p4_error: : -1
p0_31771: p4_error: : -1
p0_31772: p4_error: : -1
p0_31769: p4_error: : -1
p0_27379: p4_error: : -1
p0_27381: p4_error: : -1
p0_27383: p4_error: : -1
p0_27378: p4_error: : -1
p0_27380: p4_error: : -1
p0_27384: p4_error: : -1
p0_18051: p4_error: : -1
p0_18056: p4_error: : -1
p0_18057: p4_error: : -1
p0_18052: p4_error: : -1
p0_18054: p4_error: : -1
p0_18055: p4_error: : -1
p0_18053: p4_error: : -1
p0_18058: p4_error: : -1
p0_28550: p4_error: : -1
p0_31773: p4_error: : -1
p0_27382: p4_error: : -1
p0_27385: p4_error: : -1
cleanup

 

[eaton]

This configuration works for me. Here is the section from my logfile that matches the section you posted:

Domain Information

  Horizontal domain: nx =    144
                     ny =     96
  No. of categories: nc =      1
  No. of ice layers: ni =      4
  No. of snow layers:ns =      1
  Processors:  total    =      6
  Processor shape:       square-pop
  Distribution type:      cartesian
  Distribution weight:     latitude
  max_blocks =                 1
  Number of ghost cells:       1

CalcWorkPerBlock: Total blocks:     6 Ice blocks:     6 IceFree blocks:     0 Land blocks:     0
  Processors (X x Y) =    6 x    1
 Active processors:             6

The difference seems to be that your output is showing only 1 active processor, not 6. So it appears that the mpi job launcher has not started the job with 6 tasks.

The very top of the logfile contains information about how the components are assigned to tasks. You should see something like this for a 6 task run:

(seq_comm_setcomm)  initialize ID (  7 GLOBAL ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)
(seq_comm_setcomm)  initialize ID (  2   ATM  ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)
(seq_comm_setcomm)  initialize ID (  1   LND  ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)
(seq_comm_setcomm)  initialize ID (  4   ICE  ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)
(seq_comm_setcomm)  initialize ID (  5   GLC  ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)
(seq_comm_setcomm)  initialize ID (  3   OCN  ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)
(seq_comm_setcomm)  initialize ID (  6   CPL  ) pelist   =     0     5     1 ( npes =     6) ( nthreads =  1)

This output shows that all components are running on all 6 tasks (npes=6 means the same thing as 6 tasks).