Opened 9 years ago

Closed 9 years ago

#782 closed error (fixed)

Nested runs won't work on Phase 3

Reported by: cbirch Owned by: willie
Component: UM Model Keywords: phase 3, optimisation, cores per node
Cc: Platform:
UM Version: 7.3



I've been trying to get my vn7.3 global-12km-4km nested runs (xgfw) to work on Phase 3. I've followed the instructions on the webpage:

The global run (xgfwb) seems to complie fine and create the .astart file but it fails when executing the model run (xgfwb000.xgfwb.d12031.t155131.leave). I'm not sure why this is.

There seem to be issues with the 12 and 4km nests also but perhaps it is better to fix the global version first.


Change History (5)

comment:1 Changed 9 years ago by willie

  • Owner changed from um_support to willie
  • Status changed from new to accepted

Hi Cathryn,

You need to replace your hector_cce_7.3_file override with hector_cce_PS22_file. This is necessary in all the PS22 jobs and reflects the special optimisations that are necessary when using the Cray compiler.



comment:2 Changed 9 years ago by cbirch

Hi Willie,

I added that override and the global model now works. The 12km nest (xgfwh) compiles but I it fails in the reconfiguration with the error:

aprun: -N cannot exceed -n


I'm not sure what this means. I couldn't find a reference to it on any other ticket or by searching the source code.


comment:3 Changed 9 years ago by willie

Hi Cathryn,

You need to change the number of processors for reconfiguration for 4x6 (= 24, for the old Phase2b) to 4x8 (=32 for the Phase 3 system).



comment:4 Changed 9 years ago by cbirch

Hi Willie,

That works and the 4km also runs so you can close this now.

Thanks for your help,

comment:5 Changed 9 years ago by willie

  • Keywords 3, optimisation, cores per node added; 3 removed
  • Resolution set to fixed
  • Status changed from accepted to closed
Note: See TracTickets for help on using tickets.