Fit to mesh take a lot longer than it used to

[ole]

I have run Okushiri using the versions 4106 and the more recent

Setting the elevation data takes a lot longer than it used to - even when caching is used. Running version 4106 without caching is still lots faster than the new version .

Output from version 4106 looks like this: ===========================================

Mesh: Building neigbour structure
Mesh: Building surrogate neigbour structure
Mesh: Building boundary dictionary
Mesh: Building tagged elements dictionary
Mesh: Done
Initialising Domain
Domain: Set up communication buffers (parallel)
Domain: Initialising quantity values
Domain: Done
------------------------------------------------
Mesh statistics:
  Number of triangles = 56479
  Extent [m]:
    x in [0.000000, 5.448000]
    y in [0.000000, 3.402000]
  Areas [m^2]:
    A in [0.000004, 0.096541]
    number of distinct areas: 56479
    Histogram:
      [0.000004, 0.009657[: 56246
      [0.009657, 0.019311[: 55
      [0.019311, 0.028965[: 23
      [0.028965, 0.038619[: 9
      [0.038619, 0.048272[: 26
      [0.048272, 0.057926[: 32
      [0.057926, 0.067580[: 24
      [0.067580, 0.077234[: 32
      [0.077234, 0.086887[: 16
      [0.086887, 0.096541]: 16
    Percentiles (10%):
      5647 triangles in [0.000004, 0.000010]
      5647 triangles in [0.000010, 0.000012]
      5647 triangles in [0.000012, 0.000014]
      5647 triangles in [0.000014, 0.000016]
      5647 triangles in [0.000016, 0.000080]
      5647 triangles in [0.000080, 0.000112]
      5647 triangles in [0.000112, 0.000141]
      5647 triangles in [0.000141, 0.000190]
      5647 triangles in [0.000190, 0.000320]
      5647 triangles in [0.000320, 0.089484]
      9 triangles in [0.089484, 0.096541]
  Boundary:
    Number of boundary segments == 99
    Boundary tags == ['wall', 'wave']
------------------------------------------------

Using attribute attribute from file Benchmark_2_Bathymetry.pts
Available attributes: ['attribute']
Caching: looking for cached files /home/ole/.python_cache/fit_to_mesh[-47538529]_{Result,Args,Admin}.z
+----------------------------------------------------------
| Fri Feb  2 22:25:59 2007. Caching statistics (retrieving)
+----------------------------------------------------------
| Function:     fit_to_mesh
| Arguments:    (Array: (28290, 2), Array: (56479, 3), Array: (95892, 2), Array: (95892,))
| Keyword Args: {'alpha': 0.02, 'mesh_origin': (-1, 0.0, 0.0), 'verbose': True, 'data_origin': (-1, 0.0, 0.0)}
| CPU time:     656.89 seconds
| Loading time: 0.0 seconds
| Time saved:   656.89 seconds
|
| Caching dir:  /home/ole/.python_cache/
| Result file:  fit_to_mesh[-47538529]_Result (226372 bytes)
| Args file:    fit_to_mesh[-47538529]_Args (3432060 bytes)
| Admin file:   fit_to_mesh[-47538529]_Admin (1564 bytes)
|
| No dependencies
+----------------------------------------------------------

Reading Benchmark_2_input.tms
File_function data obtained from: Benchmark_2_input.tms
  References:
    Start time:   0.000000

====================================================

Where's the new version with blocking takes forever and looks like this

===================================================

Mesh: Building neigbour structure
Mesh: Building surrogate neigbour structure
Mesh: Building boundary dictionary
Mesh: Building tagged elements dictionary
Mesh: Done
Initialising Domain
Domain: Set up communication buffers (parallel)
Domain: Initialising quantity values
Domain: Done
------------------------------------------------
Mesh statistics:
  Number of triangles = 56541
  Extent [m]:
    x in [0.000000, 5.448000]
    y in [0.000000, 3.402000]
  Areas [m^2]:
    A in [0.000007, 0.099799]
    number of distinct areas: 56541
    Histogram:
      [0.000007, 0.009986[: 56321
      [0.009986, 0.019965[: 46
      [0.019965, 0.029945[: 21
      [0.029945, 0.039924[: 7
      [0.039924, 0.049903[: 25
      [0.049903, 0.059882[: 38
      [0.059882, 0.069861[: 32
      [0.069861, 0.079841[: 27
      [0.079841, 0.089820[: 16
      [0.089820, 0.099799]: 8
    Percentiles (10%):
      5654 triangles in [0.000007, 0.000010]
      5654 triangles in [0.000010, 0.000012]
      5654 triangles in [0.000012, 0.000014]
      5654 triangles in [0.000014, 0.000016]
      5654 triangles in [0.000016, 0.000080]
      5654 triangles in [0.000080, 0.000112]
      5654 triangles in [0.000112, 0.000142]
      5654 triangles in [0.000142, 0.000191]
      5654 triangles in [0.000191, 0.000319]
      5654 triangles in [0.000319, 0.099799]
      1 triangles in [0.099799, 0.099799]
  Boundary:
    Number of boundary segments == 95
    Boundary tags == ['wall', 'wave']
------------------------------------------------

Caching: looking for cached files c:\.python_cache\_fit[-2051356540]_{Result,Args,Admin}.z
+----------------------------------------------------------
| Fri Feb 02 14:45:01 2007. Caching statistics (retrieving)
+----------------------------------------------------------
| Function:     _fit
| Arguments:    (Array: (28319, 2), Array: (56541, 3))
| Keyword Args: {'alpha': 0.02, 'mesh_origin': None, 'verbose': True}
| CPU time:     137.47 seconds
| Loading time: 1.64 seconds
| Time saved:   135.83 seconds
|
| Caching dir:  c:\.python_cache\
| Result file:  _fit[-2051356540]_Result.z (11208238 bytes, compressed)
| Args file:    _fit[-2051356540]_Args.z (712707 bytes, compressed)
| Admin file:   _fit[-2051356540]_Admin.z (167 bytes, compressed)
|
| No dependencies
+----------------------------------------------------------

Got 1 variables: ['attribute']
Doing 0 of 95902
Block 0
Reading Benchmark_2_input.tms
File_function data obtained from: Benchmark_2_input.tms
  References:
    Start time:   0.000000
Time = 0.0000, steps=0 (0)

===================================================

At this point 'Block 0' is was hanging for more than 30 minutes while I was able to run version 4106 twice both with and without caching.

[duncan]

This is not occuring on cyclone and on my PC. 4106 is not constanly faster on Nautilus.

So I'm closing this ticket as a works for me.