[GRASS-dev] Week 11: GRASS GIS 3D flowlines

GRASS GIS GRASS GIS developer
1

Hi,

  1. What did you get done this week:
    I implemented adding attributes to each segment of a flowline. I added column with velocities and values of input 3D raster along the flowline. The result can be visualized by using vector color tables (v.colors) in 2D and 3D.
    I started writing r3.gradient module based on the gradient function in r3.flow. The module is already running and giving correct results, I will add it to addons once the gradient function is moved to library.

  2. What do you plan on doing next week?
    I will try to parallelize the gradient computation with openMP, the structure of the code is already prepared for that. Then, I plan to add more tests.

  3. Are you blocked on anything?
    No.

Anna

[1] http://trac.osgeo.org/grass/wiki/GSoC/2014/ImplementationOf3DRasterFlowLine
[2] http://trac.osgeo.org/grass/browser/grass-addons/grass7/raster3d/r3.flow

2

Anna,
Will you post the benchmarking results for the OpenMP computation vs the non-OpenMP?

Doug

···

On Fri, Aug 1, 2014 at 3:54 PM, Anna Petrášová <kratochanna@gmail.com> wrote:

Hi,

  1. What did you get done this week:
    I implemented adding attributes to each segment of a flowline. I added column with velocities and values of input 3D raster along the flowline. The result can be visualized by using vector color tables (v.colors) in 2D and 3D.
    I started writing r3.gradient module based on the gradient function in r3.flow. The module is already running and giving correct results, I will add it to addons once the gradient function is moved to library.

  2. What do you plan on doing next week?
    I will try to parallelize the gradient computation with openMP, the structure of the code is already prepared for that. Then, I plan to add more tests.

  3. Are you blocked on anything?
    No.

Anna

[1] http://trac.osgeo.org/grass/wiki/GSoC/2014/ImplementationOf3DRasterFlowLine
[2] http://trac.osgeo.org/grass/browser/grass-addons/grass7/raster3d/r3.flow

grass-dev mailing list
grass-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-dev

–

Doug Newcomb
USFWS
Raleigh, NC
919-856-4520 ext. 14 doug_newcomb@fws.gov

The opinions I express are my own and are not representative of the official policy of the U.S.Fish and Wildlife Service or Dept. of the Interior. Life is too short for undocumented, proprietary data formats.

3

Hi,

···

On Tue, Aug 5, 2014 at 11:04 AM, Newcomb, Doug <doug_newcomb@fws.gov> wrote:

Anna,
Will you post the benchmarking results for the OpenMP computation vs the non-OpenMP?

after some testing, my conclusion is that OpenMP does not really speed up gradient computation and it can even slow it down. I tried it on my laptop with 2 cores and on a powerful computer with 12 cores, with different sizes of processed blocks and I didn’t get any significant speed-up. The most costly seems to be writing of the three 3D rasters and the gradient computation is relatively short. I also tested different cache settings but default settings seems to be optimal, I got only similar or worse results with different cache settings. I hope I didn’t missed anything, since this is quite new to me.

Best,

Anna

Doug

On Fri, Aug 1, 2014 at 3:54 PM, Anna Petrášová <kratochanna@gmail.com> wrote:

Hi,

  1. What did you get done this week:
    I implemented adding attributes to each segment of a flowline. I added column with velocities and values of input 3D raster along the flowline. The result can be visualized by using vector color tables (v.colors) in 2D and 3D.
    I started writing r3.gradient module based on the gradient function in r3.flow. The module is already running and giving correct results, I will add it to addons once the gradient function is moved to library.

  2. What do you plan on doing next week?
    I will try to parallelize the gradient computation with openMP, the structure of the code is already prepared for that. Then, I plan to add more tests.

  3. Are you blocked on anything?
    No.

Anna

[1] http://trac.osgeo.org/grass/wiki/GSoC/2014/ImplementationOf3DRasterFlowLine
[2] http://trac.osgeo.org/grass/browser/grass-addons/grass7/raster3d/r3.flow

grass-dev mailing list
grass-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-dev

–

Doug Newcomb
USFWS
Raleigh, NC
919-856-4520 ext. 14 doug_newcomb@fws.gov

The opinions I express are my own and are not representative of the official policy of the U.S.Fish and Wildlife Service or Dept. of the Interior. Life is too short for undocumented, proprietary data formats.

4

Anna,
This parallels my experience in running v.surf.rst with OpenMP enabled and disabled on large areas. With OpenMP enabled on an 8 core computer, the process used 700% of CPU time, but took about twice as long as running the same data set on a single core with OpenMP disabled.

Doug

···

On Wed, Aug 6, 2014 at 10:58 PM, Anna Petrášová <kratochanna@gmail.com> wrote:

Hi,

–

Doug Newcomb
USFWS
Raleigh, NC
919-856-4520 ext. 14 doug_newcomb@fws.gov

The opinions I express are my own and are not representative of the official policy of the U.S.Fish and Wildlife Service or Dept. of the Interior. Life is too short for undocumented, proprietary data formats.

On Tue, Aug 5, 2014 at 11:04 AM, Newcomb, Doug <doug_newcomb@fws.gov> wrote:

Anna,
Will you post the benchmarking results for the OpenMP computation vs the non-OpenMP?

after some testing, my conclusion is that OpenMP does not really speed up gradient computation and it can even slow it down. I tried it on my laptop with 2 cores and on a powerful computer with 12 cores, with different sizes of processed blocks and I didn’t get any significant speed-up. The most costly seems to be writing of the three 3D rasters and the gradient computation is relatively short. I also tested different cache settings but default settings seems to be optimal, I got only similar or worse results with different cache settings. I hope I didn’t missed anything, since this is quite new to me.

Best,

Anna

Doug

On Fri, Aug 1, 2014 at 3:54 PM, Anna Petrášová <kratochanna@gmail.com> wrote:

Hi,

  1. What did you get done this week:
    I implemented adding attributes to each segment of a flowline. I added column with velocities and values of input 3D raster along the flowline. The result can be visualized by using vector color tables (v.colors) in 2D and 3D.
    I started writing r3.gradient module based on the gradient function in r3.flow. The module is already running and giving correct results, I will add it to addons once the gradient function is moved to library.

  2. What do you plan on doing next week?
    I will try to parallelize the gradient computation with openMP, the structure of the code is already prepared for that. Then, I plan to add more tests.

  3. Are you blocked on anything?
    No.

Anna

[1] http://trac.osgeo.org/grass/wiki/GSoC/2014/ImplementationOf3DRasterFlowLine
[2] http://trac.osgeo.org/grass/browser/grass-addons/grass7/raster3d/r3.flow

grass-dev mailing list
grass-dev@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-dev

–

Doug Newcomb
USFWS
Raleigh, NC
919-856-4520 ext. 14 doug_newcomb@fws.gov

The opinions I express are my own and are not representative of the official policy of the U.S.Fish and Wildlife Service or Dept. of the Interior. Life is too short for undocumented, proprietary data formats.