Dear all,
I'd like to run the new UEMS with three domains:
1* Mother domain, including Canary Archipelago, Madeira and Azores (9 km resolution, Mercator)
2* Second domain, covering the whole Canary Archipelago (3km resolution)
3* One las domain at 1 km resolution, covering just one island.
The forecasted time would be 36-48 hours
In order to do that, I've seen this linux box (CentOS 6.7):
*********************
2x Intel® Xeon® E5-2630v3 - 8C/16T
32GB - 8x 4GB DDR4 2133 ECC REG - 2 CPUs
SSD 240GB 2.5" 550MB/s
Nvidia GTX 970 4GB GDDR5
*********************
Do you think it would be enough? Also, Is someone experienced in running WRFEMS or UEMS over CentOS 7? It uses Kernel 3.XX
Best,
Luis
Is this hardware enough?
Re: Is this hardware enough?
Hi Icana,
I have a closely similar machine to yours:
2 x AMD Opteron 6276 with 8 physical cores each (16 threads each),
64 GB memory
500 GB SSD.
I tried the domain you suggested and it does seem a bit on the heavy side for the suggested machine (see namelist.wps below).
There are > 200 x 200 cells for each domain level.
I ran all d1,d2,d3 for 3 hour using the new UEMS-ARW core with Adaptive timing and it took 27min 31sec.
That would need about 4 hours for a 24 hour forecast.
I suggest to try 12km dx for the top domain and 4km for d2 with 1.33km for d3. I think that it would put you in much more reasonable timing without significant reduction in the quality of the forecast
David
I have a closely similar machine to yours:
2 x AMD Opteron 6276 with 8 physical cores each (16 threads each),
64 GB memory
500 GB SSD.
I tried the domain you suggested and it does seem a bit on the heavy side for the suggested machine (see namelist.wps below).
There are > 200 x 200 cells for each domain level.
Code: Select all
&share
wrf_core = 'ARW'
max_dom = 3
start_date = '2016-01-30_12:00:00', '2016-01-30_12:00:00', '2016-01-30_12:00:00'
end_date = '2016-01-30_15:00:00', '2016-01-30_12:00:00', '2016-01-30_12:00:00'
interval_seconds = 10800
io_form_geogrid = 2
opt_output_from_geogrid_path = '/home/david/uems/runs/test/static/'
debug_level = 0
/
&geogrid
parent_id = 1, 1, 2
parent_grid_ratio = 1, 3, 3
i_parent_start = 1, 159, 66
j_parent_start = 1, 47, 74
e_we = 262, 214, 199
e_sn = 267, 211, 199
geog_data_res = 'modis_30s+5m', 'modis_30s+30s', 'modis_30s+30s'
dx = 9000
dy = 9000
map_proj = 'mercator'
ref_lat = 32.628
ref_lon = -21.927
truelat1 = 32.628
truelat2 = 0
stand_lon = -21.927
ref_x = 131
ref_y = 133.5
geog_data_path = '/home/david/uems/data/geog'
opt_geogrid_tbl_path = '/home/david/uems/runs/test/static/'
/
&ungrib
out_format = 'WPS'
prefix = 'FILE'
/
&metgrid
fg_name = GFSP25
io_form_metgrid = 2
opt_output_from_metgrid_path = '/home/david/uems/runs/test/wpsprd'
opt_metgrid_tbl_path = '/home/david/uems/runs/test'
process_only_bdy =
/
&mod_levs
press_pa = 201300, 200100, 100000, 95000, 90000, 85000, 80000, 75000, 70000, 65000, 60000, 55000, 50000, 45000, 40000, 35000, 30000, 25000, 20000, 15000, 10000, 5000, 1000
/
&domain_wizard
grib_data_path = 'null'
grib_vtable = 'Vtable.GFS'
dwiz_center_over_gmt = true
dwiz_desc = test
dwiz_gridpt_dist_km = 9.0
dwiz_latlon_linecolor = -8355712
dwiz_latlon_space_in_deg = 10
dwiz_map_horiz_scrollbar_pos = 3259
dwiz_map_scale_pct = 50.0
dwiz_map_vert_scrollbar_pos = 994
dwiz_mpi_command = null
dwiz_name = test
dwiz_show_political = true
dwiz_user_rect_x1 = 3314
dwiz_user_rect_x2 = 3800
dwiz_user_rect_y1 = 1083
dwiz_user_rect_y2 = 1499
dwiz_modis = true
dwiz_tcvitals = null
dwiz_bigmap = Y
dwiz_lakes = false
/
That would need about 4 hours for a 24 hour forecast.
I suggest to try 12km dx for the top domain and 4km for d2 with 1.33km for d3. I think that it would put you in much more reasonable timing without significant reduction in the quality of the forecast
David
Re: Is this hardware enough?
Thank you very much David. Your suggestion is very much appreciated.
I'll follow your advice
Luis
I'll follow your advice
Luis
Who is online
Users browsing this forum: No registered users and 4 guests