Changes between Version 658 and Version 659 of doc/app/errmsg

Timestamp:

Oct 15, 2019 5:03:11 PM (5 years ago)

Author:

scharf

Comment:

--

doc/app/errmsg

@@ -266 +266 @@
 ||[=#PA0223 PA0223]   ||errors in \$nesting_parameters   ||An unvalid variable name or value has been given in the nestpar-NAMELIST. Carefully check your nestpar-NAMELIST and compare with the parameters given and explained in the [wiki:doc/app/nestpar Nesting]-section of the model steering parameters.  ||
 ||[=#PA0224 PA0224]   || Eeach dz_stretch_level_end has to be ... than its corresponding value for dz_stretch_level_start ... 4*MAX(dz(n),dz(n+1)) to allow for smooth grid stretching  || To allow for a smooth stretching, the distance between [../inipar#dz_stretch_level_start dz_stretch_level_start] and [../inipar#dz_stretch_level_end dz_stretch_level_end] has to be large enough. Please increase the stretching region appropriately. ||
-||[=#PA0225 PA0225]   ||x-direction: gridpoint number (...) is not an integral divisor of the number of processors (...)   || The division of the number of grid points along the x-direction given by [../inipar#nx nx]+1 by the number of processors along the x-direction has a rest and is therefore not an integral divisor of the number of processors. To solve this problem you must ensure that the aforementioned division is without rest for example by setting the number of PEs in x-direction manually (see [../d3par#npex npex]) or by changing [../inipar#nx nx]. It might also be the case that you did not think about that [../inipar#nx nx]+1 is the total number of gridpoints along the x-direction instead of [../inipar#nx nx]. ||
+||[=#PA0225 PA0225]   ||x-direction: gridpoint number (...) is not an integral multiple of the number of processors (...)   || The division of the number of grid points along the x-direction given by [../inipar#nx nx]+1 by the number of processors along the x-direction has a rest and is therefore not an integral multiple of the number of processors. To solve this problem you must ensure that the aforementioned division is without rest for example by setting the number of PEs in x-direction manually (see [../d3par#npex npex]) or by changing [../inipar#nx nx]. It might also be the case that you did not think about that [../inipar#nx nx]+1 is the total number of gridpoints along the x-direction instead of [../inipar#nx nx]. ||
 ||[=#PA0226 PA0226]   || Eeach dz_stretch_level_start has to be ... than ... || The stretching region has to be defined above the first two grid points (Prandtl-layer). In case of ocean mode the stretching region muss below the two highest grid points (surface). Please adjust [../inipar#dz_stretch_level_start dz_stretch_level_start] accordingly. ||
-||[=#PA0227 PA0227]   ||y-direction: gridpoint number (...) is not an integral divisor of the number of processors (...)   || The division of the number of grid points along the y-direction given by [../inipar#ny ny]+1 by the number of processors along the y-direction has a rest and is therefore not an integral divisor of the number of processors. To solve this problem you must ensure that the aforementioned division is without rest for example by setting the number of PEs in y-direction manually (see [../d3par#npey npey]) or by changing [../inipar#ny ny]. It might also be the case that you did not think about that [../inipar#ny ny]+1 is the total number of gridpoints along the y-direction instead of [../inipar#ny ny]. ||
+||[=#PA0227 PA0227]   ||y-direction: gridpoint number (...) is not an integral multiple of the number of processors (...)   || The division of the number of grid points along the y-direction given by [../inipar#ny ny]+1 by the number of processors along the y-direction has a rest and is therefore not an integral multiple of the number of processors. To solve this problem you must ensure that the aforementioned division is without rest for example by setting the number of PEs in y-direction manually (see [../d3par#npey npey]) or by changing [../inipar#ny ny]. It might also be the case that you did not think about that [../inipar#ny ny]+1 is the total number of gridpoints along the y-direction instead of [../inipar#ny ny]. ||
 ||[=#PA0228 PA0228]   || Two adjacent values of dz must be different || It is necessary to specify different values for [../inipar#dz dz] if they are adjacent. ||
 ||[=#PA0229 PA0229]   ||nesting-setup requires different number of MPI procs (...) than provided (...)   ||This message appears for nested runs, if the total number of MPI processes for all nested domains given in the nestpar-NAMELIST is larger than the number of MPI processes which has been allowed for the run, e.g. with palmrun-option -X. (Attention: For hybrid OpenMP/MPI-runs the number of MPI processes allowed for the run is the number given by -X divided by the number of threads per MPI-task, as given with palmrun-option -O.) You either have to provide the run more resources (i.e. more cores) or you have to decrease the number of MPI processes in $nestpar. ||
-||[=#PA0230 PA0230]   ||transposition z --> x: nz=... is not an integral divisor of pdims(1)= ...   ||Due to restrictions in data transposition, the division of [../inipar#nz nz] by the number of processors in x-direction has to be without rest. Please ensure this by changing [../inipar#nz nz] or setting the number of gridpoints in x-direction (pdims(1)) manually by [../d3par#npex npex].  ||
-||[=#PA0231 PA0231]   ||transposition x --> y: nx+1=... is not an integral divisor of pdims(2)= ...   || Due to restrictions in data transposition, the division of [../inipar#nx nx]+1 by the number of processors in y-direction has to be without rest. Please ensure this by changing [../inipar#nx nx] or setting the number of gridpoints in y-direction (pdims(2)) manually by [../d3par#npey npex]. ||
-||[=#PA0232 PA0232]   ||transposition y --> z: ny+1=... is not an integral divisor of pdims(1)=...    ||Due to restrictions in data transposition, the division of [../inipar#ny ny]+1 by the number of processors in x-direction has to be without rest. Please ensure this by changing [../inipar#ny ny] or setting the number of gridpoints in x-direction (pdims(1)) manually by [../d3par#npex npex].  ||
-||[=#PA0233 PA0233]   ||transposition x --> y: ny+1=... is not an integral divisor of pdims(1)=...   ||Due to restrictions in data transposition, the division of [../inipar#ny ny]+1 by the number of processors in x-direction has to be without rest. Please ensure this by changing [../inipar#ny ny] or setting the number of gridpoints in x-direction (pdims(1)) manually by [../d3par#npex npex].  ||
-||[=#PA0234 PA0234]   ||direct transposition z --> y (needed for spectra): nz=... is not an integral divisor of pdims(2)=...   || Due to restrictions in data transposition, the division of [../inipar#nz nz] by the number of processors in y-direction has to be without rest. Please ensure this by changing [../inipar#nz nz] or setting the number of gridpoints in y-direction (pdims(2)) manually by [../d3par#npey npey]. ||
+||[=#PA0230 PA0230]   ||transposition z --> x: nz=... is not an integral multiple of pdims(1)= ...   ||Due to restrictions in data transposition, the division of [../inipar#nz nz] by the number of processors in x-direction has to be without rest. Please ensure this by changing [../inipar#nz nz] or setting the number of gridpoints in x-direction (pdims(1)) manually by [../d3par#npex npex].  ||
+||[=#PA0231 PA0231]   ||transposition x --> y: nx+1=... is not an integral multiple of pdims(2)= ...   || Due to restrictions in data transposition, the division of [../inipar#nx nx]+1 by the number of processors in y-direction has to be without rest. Please ensure this by changing [../inipar#nx nx] or setting the number of gridpoints in y-direction (pdims(2)) manually by [../d3par#npey npex]. ||
+||[=#PA0232 PA0232]   ||transposition y --> z: ny+1=... is not an integral multiple of pdims(1)=...    ||Due to restrictions in data transposition, the division of [../inipar#ny ny]+1 by the number of processors in x-direction has to be without rest. Please ensure this by changing [../inipar#ny ny] or setting the number of gridpoints in x-direction (pdims(1)) manually by [../d3par#npex npex].  ||
+||[=#PA0233 PA0233]   ||transposition x --> y: ny+1=... is not an integral multiple of pdims(1)=...   ||Due to restrictions in data transposition, the division of [../inipar#ny ny]+1 by the number of processors in x-direction has to be without rest. Please ensure this by changing [../inipar#ny ny] or setting the number of gridpoints in x-direction (pdims(1)) manually by [../d3par#npex npex].  ||
+||[=#PA0234 PA0234]   ||direct transposition z --> y (needed for spectra): nz=... is not an integral multiple of pdims(2)=...   || Due to restrictions in data transposition, the division of [../inipar#nz nz] by the number of processors in y-direction has to be without rest. Please ensure this by changing [../inipar#nz nz] or setting the number of gridpoints in y-direction (pdims(2)) manually by [../d3par#npey npey]. ||
 ||[=#PA0235 PA0235]   ||mg_switch_to_pe0_level out of range and reset to 0  || mg_switch_to_pe0_level was manually set out of range. It was reset to 0, so that the gathering level is determined automatically and displayed in file RUN_CONTROL. See [../inipar#mg_switch_to_pe0_level mg_switch_to_pe0_level] for details. ||
 ||[=#PA0236 PA0236]   ||grid coarsening on subdomain level cannot be performed  || The multigrid pressure solver halfs the number of grid points in each direction to get coarser grid levels. This is done for the subdomains on each PE until one of the directions cannot be divided by 2 without rest. If this is the case, the coarsest grid level of each subdomain are gathered on each PE for further coarsening. Before gathering, the coarsening must be done for the subdomain at least one time. You get this error messages for example with the following combination: \\ [../inipar#nx  nx ] = 60, [../inipar#ny  ny ] = 60, [../inipar#nz  nz ] = 60, \\ [../d3par#npex  npex ] = 4, [../d3par#npey  npey ] = 2. \\ With this combination, the subdomain on each PE has the size: \\ nx = 15, ny = 30, nz = 60. \\ The number of grid points in the x-direction of the subdomain cannot be divided by 2 and no coarsening is possible for the subdomains in this example. \\\\ To avoid this error ensure that the number of grid points in each direction can be divided by 2 without rest on the subdomain of an PE at least one time. Therefor vary the number of grid points for the direction which does not fit by changing [../inipar#nx  nx ], [../inipar#ny  ny ] or [../inipar#nz  nz ] or adjust the number of processors by modifying the parameters [../d3par#npex  npex ] or [../d3par#npey  npey ] in order to get other subdomain sizes. \\ In the example given above, you can avoid the error by changing nx or npex. For example setting [../inipar#nx  nx ] = 64 leads to a subdomain of \\ nx = 16, ny = 30, nz = 60. \\ All numbers are now divisible by 2 without rest and the first coarser grid on the subdomain is then: \\ nx = 8, ny = 15, nz = 30. ||