Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

grid dimensions

federico

New member
Dear support,

I need to produce a WRF simulation at very hight horizontal resolution for a Medicane. All the programs are fine until wrf.exe. When I run wrf.exe the model exit before writing the first wrfout file.

I have the following error:
*************************************
Nesting domain
ids,ide,jds,jde 1 1041 1 801
ims,ime,jms,jme -4 184 -4 144
ips,ipe,jps,jpe 1 174 1 134
INTERMEDIATE domain
ids,ide,jds,jde 398 663 148 353
ims,ime,jms,jme 393 453 143 193
ips,ipe,jps,jpe 396 443 146 183
*************************************
d01 2020-09-16_12:00:00 alloc_space_field: domain 2 , 26866596 bytes allocated
d01 2020-09-16_12:00:00 alloc_space_field: domain 2 , 802921876 bytes allocated
d01 2020-09-16_12:00:00 *** Initializing nest domain # 2 from an input file. ***
d01 2020-09-16_12:00:00 med_initialdata_input: calling input_input
d01 2020-09-16_12:00:00 Input data is acceptable to use: wrfinput_d02

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x7FD05085D6D7
#1 0x7FD05085DD1E
#2 0x7FD04FD583FF
#3 0x7FD04FE76CB5
#4 0x307EFB5 in MPID_Segment_contig_m2m
#5 0x307E90B in MPID_Segment_manipulate
#6 0x30855B9 in MPID_Segment_pack
#7 0x3073EB3 in lmt_shm_send_progress at mpid_nem_lmt_shm.c:?
#8 0x30750B3 in MPID_nem_lmt_shm_start_send
#9 0x3072F5B in pkt_CTS_handler at mpid_nem_lmt.c:?
#10 0x306D3AC in MPIDI_CH3I_Progress
#11 0x30B3EA8 in MPIR_Waitall_impl
#12 0x304B824 in MPIC_Waitall_ft
#13 0x3037DB4 in MPIR_Scatterv




Now I tried two different configurations:

e_we = 800, 1041, 94,
e_sn = 800, 801, 91,
e_vert = 45, 45, 33,
....
dx = 2400, 600, 3333.33,
dy = 2400, 600, 3333.33,
grid_id = 1, 2, 3,
parent_id = 0, 1, 2,
i_parent_start = 1, 400, 30,
j_parent_start = 1, 150, 30,
parent_grid_ratio = 1, 4, 3,
parent_time_step_ratio = 1, 4, 3,
feedback = 1,
smooth_option = 0

With this configuration the model doesn't work (previous error). However it works fine with the following:

e_we = 800, 1021, 94,
e_sn = 800, 721, 91,
e_vert = 55, 55, 33,
....
dx = 2400, 600, 3333.33,
dy = 2400, 600, 3333.33,
grid_id = 1, 2, 3,
parent_id = 0, 1, 2,
i_parent_start = 1, 400, 30,
j_parent_start = 1, 150, 30,
parent_grid_ratio = 1, 4, 3,
parent_time_step_ratio = 1, 4, 3,
feedback = 1,
smooth_option = 0

Is there any maximum dimension in the e_we, e_sn or in the number e_we*e_sn*e_vert that I should respect? Can you give me some hints?

Thanks.

Stefano
 
Stefano,
This is possibly a memory issue since the job failed immediately with segmentation fault. Can you run this case with more processors (which will give you larger memory)?
 
Top