新手求助！./wrf.exe太慢，尝试mpi并行错误

不二兔

单独运行./wrf.exe没有错误，但是时间太长，想尝试mpi并行运算，但是一直不成功，请问需要修改哪里呢？
mpirun -np 4 ./wrf.exe
starting wrf task            0  of            4
starting wrf task            1  of            4
starting wrf task            3  of            4
starting wrf task            2  of            4
rank 3 in job 8  localhost.localdomain_32858   caused collective abort of all ranks exit status of rank 3: killed by signal 11

编译的时候选的是并行，rsl.out.0000中也没有明显的错误。
以下是rsl.out.0000：
taskid: 0 hostname: localhost.localdomain
Quilting with   1 groups of   0 I/O tasks.
Ntasks in X            2, ntasks in Y            2
--- WARNING: traj_opt is zero, but num_traj is not zero; setting num_traj to zero.
--- NOTE: sst_update is 0, setting io_form_auxinput4 = 0 and auxinput4_interval = 0 for all domains
--- NOTE: sst_update is 0, setting io_form_auxinput4 = 0 and auxinput4_interval = 0 for all domains
--- NOTE: grid_fdda is 0 for domain      1, setting gfdda interval and ending time to 0 for that domain.
--- NOTE: both grid_sfdda and pxlsm_soil_nudge are 0 for domain      1, setting sgfdda interval and ending time to 0 for that domain.
--- NOTE: obs_nudge_opt is 0 for domain      1, setting obs nudging interval and ending time to 0 for that domain.
--- NOTE: grid_fdda is 0 for domain      2, setting gfdda interval and ending time to 0 for that domain.
--- NOTE: both grid_sfdda and pxlsm_soil_nudge are 0 for domain      2, setting sgfdda interval and ending time to 0 for that domain.
--- NOTE: obs_nudge_opt is 0 for domain      2, setting obs nudging interval and ending time to 0 for that domain.
--- NOTE: bl_pbl_physics /= 4, implies mfshconv must be 0, resetting
Need MYNN PBL for icloud_bl = 1, resetting to 0
--- NOTE: RRTMG radiation is not used, setting:  o3input=0 to avoid data pre-processing
*************************************
Configuring physics suite 'conus'

         mp_physics:      8      8
         cu_physics:      6      6
      ra_lw_physics:      4      4
      ra_sw_physics:      4      4
     bl_pbl_physics:      2      2
  sf_sfclay_physics:      2      2
sf_surface_physics:      2      2
*************************************
--- NOTE: RRTMG radiation is in use, setting:  levsiz=59, alevsiz=12, no_src_types=6
--- NOTE: num_soil_layers has been set to      4
WRF V3.9 MODEL
*************************************
Parent domain
ids,ide,jds,jde            1          74           1          61
ims,ime,jms,jme           -4          44          -4          37
ips,ipe,jps,jpe            1          37           1          30
*************************************
DYNAMICS OPTION: Eulerian Mass Coordinate
   alloc_space_field: domain            1,               48983984 bytes allocated
  med_initialdata_input: calling input_input
Timing for processing wrfinput file (stream 0) for domain        1:    0.06245 elapsed seconds
Max map factor in domain 1 =  1.05. Scale the dt in the model accordingly.
INPUT LandUse = "MODIFIED_IGBP_MODIS_NOAH"
LANDUSE TYPE = "MODIFIED_IGBP_MODIS_NOAH" FOUND          33 CATEGORIES           2 SEASONS WATER CATEGORY =           17 SNOW CATEGORY =           15
INITIALIZE THREE Noah LSM RELATED TABLES
Skipping over LUTYPE = USGS
LANDUSE TYPE = MODIFIED_IGBP_MODIS_NOAH FOUND          20 CATEGORIES
INPUT SOIL TEXTURE CLASSIFICATION = STAS
SOIL TEXTURE CLASSIFICATION = STAS FOUND          19 CATEGORIES
ThompMP: read qr_acr_qg.dat stead of computing
ThompMP: read qr_acr_qs.dat instead of computing
ThompMP: read freezeH2O.dat stead of computing
*************************************
Nesting domain
ids,ide,jds,jde            1         112           1          97
ims,ime,jms,jme           -4          66          -4          60
ips,ipe,jps,jpe            1          56           1          48
INTERMEDIATE domain
ids,ide,jds,jde           29          71          15          52
ims,ime,jms,jme           24          59          10          42
ips,ipe,jps,jpe           27          49          13          32
*************************************
d01 2013-12-29_00:00:00  alloc_space_field: domain            2,               11129184 bytes allocated
d01 2013-12-29_00:00:00  alloc_space_field: domain            2,              106361276 bytes allocated
d01 2013-12-29_00:00:00 *** Initializing nest domain # 2 from an input file. ***
d01 2013-12-29_00:00:00 med_initialdata_input: calling input_input
Timing for processingwrfinput file (stream 0) for domain        2:    0.17978 elapsed secondsINPUT LandUse = "MODIFIED_IGBP_MODIS_NOAH"
LANDUSE TYPE = "MODIFIED_IGBP_MODIS_NOAH" FOUND          33 CATEGORIES           2 SEASONS WATER CATEGORY =           17 SNOW CATEGORY =           15
INITIALIZE THREE Noah LSM RELATED TABLES
Skipping over LUTYPE = USGS
LANDUSE TYPE = MODIFIED_IGBP_MODIS_NOAH FOUND          20 CATEGORIES
INPUT SOIL TEXTURE CLASSIFICATION = STAS
SOIL TEXTURE CLASSIFICATION = STAS FOUND          19 CATEGORIES
INPUT LandUse = "MODIFIED_IGBP_MODIS_NOAH"
LANDUSE TYPE = "MODIFIED_IGBP_MODIS_NOAH" FOUND          33 CATEGORIES           2 SEASONS WATER CATEGORY =           17 SNOW CATEGORY =           15
INITIALIZE THREE Noah LSM RELATED TABLES
Skipping over LUTYPE = USGS
LANDUSE TYPE = MODIFIED_IGBP_MODIS_NOAH FOUND          20 CATEGORIES
INPUT SOIL TEXTURE CLASSIFICATION = STAS
SOIL TEXTURE CLASSIFICATION = STAS FOUND          19 CATEGORIES
Max map factor in domain 1 =  1.05. Scale the dt in the model accordingly.
INPUT LandUse = "MODIFIED_IGBP_MODIS_NOAH"
LANDUSE TYPE = "MODIFIED_IGBP_MODIS_NOAH" FOUND          33 CATEGORIES           2 SEASONS WATER CATEGORY =           17 SNOW CATEGORY =           15
INITIALIZE THREE Noah LSM RELATED TABLES
Skipping over LUTYPE = USGS
LANDUSE TYPE = MODIFIED_IGBP_MODIS_NOAH FOUND          20 CATEGORIES
INPUT SOIL TEXTURE CLASSIFICATION = STAS
SOIL TEXTURE CLASSIFICATION = STAS FOUND          19 CATEGORIES
Timing for Writing wrfout_d01_2013-12-29_00:00:00 for domain        1:    0.34781 elapsed seconds
Timing for processing lateral boundary for domain        1:    0.02499 elapsed seconds

xinshou6

那0003有错吗 都看看呢 serial没毛病呗

不二兔

串行可以运行，rsl.out.000*这几个中都没有明显的错误。。

不二兔

xinshou6 发表于 2020-11-25 11:09
那0003有错吗 都看看呢 serial没毛病呗

串行可以运行，rsl.out.000*这几个中都没有明显的错误http://bbs.06climate.com/forum.p ... 6&highlight=mpi这个帖子和我的问题一样，可是我还是不太会怎么解决。。

不二兔

网上说段错误是内存原因，尝试清理一下缓存和内存，再加上ulimit -s unlimited 并行就可以用了。