T85_gx1v3 run in Cray X1E

[songbg@apcc21_net]

I am trying to run CCSM3 with T85_gx1v3 B dataset in Cray X1E.
T42_gx1v3 B run was successful.
For T85, the compiling was Ok but running has some problem.
Could you give me some comments or suggestions to solve the problem?

Thanks,

Bong-Gen Song

The run log is:
.
.
.
(main) =========================================================================
(main) CCSM Coupler, version 6 (cpl6)
(main) CVS tag $Name$
(main) date & time: 2008-01-24 01:57:54
(main) =========================================================================
(cpl_comm_init) setting up communicators, name = cpl
===================================
(cpl_comm_init) cpl_comm_comp, size: 5 2
(cpl_comm_init) comm world : comm,npe,pid 4 56 1
(cpl_comm_init) comm component: comm,npe,pid 5 2 1
(cpl_comm_init) comm world pe0: atm,ice,lnd,ocn,cpl,me 40 2 10 16 0 0
(cpl_comm_init) mph cid : atm,ice,lnd,ocn,cpl,me 1 2 3 4 5 5
(shr_msg_chdir) read cpl_stdio.nml, changed cwd to /san_home/users/songbg/nohara/CCSM3/CCSM.T85.TEST/cpl
(shr_msg_chdir) read ice_stdio.nml, changed cwd to /san_home/users/songbg/nohara/CCSM3/CCSM.T85.TEST/ice
(shr_msg_chStdIn) read cpl_stdio.nml, unit 5 connected to cpl.stdin
(shr_msg_chStdIn) read ice_stdio.nml, unit 5 connected to ice.stdin
sh: msread: not found

Traceback for process 14745435(ssp mode) apid 14745435.0 on node 0
__open+0x018C (0x1351194) at open.c
_open+0x0290 (0x13517F8) at open.c:58
_do_open+0x0240 (0x14AC3E0) at fopn.c:983
_f_opn+0x07B0 (0x14AAD50) at fopn.c:434
_f_open+0x1A6C (0x13A2EEC) at open.c:620
__OPN+0x0C68 (0x1229CC0) at opn.c:389
_OPEN+0x00D0 (0x122A068) at opn.c:426
shr_msg_chstdout@shr_msg_mod_+0x0928 (0x116A8C8) at shr_msg_mod.f90:255
shr_msg_dirio@shr_msg_mod_+0x00D4 (0x11685D4) at shr_msg_mod.f90:148
cpl_+0x14A8 (0x110F008) at main.f90:174
Fault: memory fault on accelerated page: 0xE0FF5B
Segmentation fault (core dumped)

 

[songbg@apcc21_net]

Hi,

I solved the problem.
Well, the Cray X1E I used is not "Phoenix" machine.
I use KMA/Cray X1E (Korean Meteorological Administration).

The problem were;

1. Not fully uploaded the inputdata. ^^;
So run log showed "sh: msread: not found".

2. Load balance in Cray X1E: I tried to use different CPU sets in Cray X1E.
For example,
(cpl: 4, cam: 32, clm: 4, csim: 8, pop: 16) total 64 was OK.
(cpl: 2, cam: 32, clm: 4, csim: 8, pop: 16) total 62 was OK. (slower than the above)
(cpl: 2, cam: 40, clm: 2, csim: 4, pop: 16) total 64 didn't work. (bus error)
(cpl: 8, cam: 64, clm: 8, csim: 16, pop: 32) total 128 didn't work. (bus error)

I'm still trying to find a good combination of CPUs (more than 100 or 200) but it is difficult without using the timing program in CCSM3 scripts.

Thank you.

Bong-Geun Song