GPU Speed-up: The future of 2D Hydraulic Modeling is here!

RiverFlow2D Plus GPU is our new two-dimensional model that achieves simulation speeds never seen before allowing solving large river hydraulic problems that used to be impractical or even impossible until recently.

The release of the GPU version of RiverFlow2D Plus flexible mesh model is already creating a splash in many flood and river modeling applications, running 30 times faster or more than non-parallelized or CPU parallelized models.

Traditional CPU processor performance has reached its peak but  Hydronia’s RiverFlow2D Plus GPU offers a new high performance numerical engine, that when run using off-the-shelf NVIDIA graphics cards brings incredible performance to all modelers and helps them finalize projects in a fraction of the time.

 

RiverFlow2D Plus GPU advantages

GPU-based Finite volume computational engine

Triangular unstructured flexible mesh

Comprehensive Mesh Generation pre-processor and Post-processor using Argus ONE

High accuracy (Zero Volume Conservation Error)

Fast computations (>100X speedup)

 

RiverFlow2D Plus GPU main capabilities

 

Mesh Generator

     •  Automatic generation of flexible

        triangular- element mesh 

     •  Unlimited element refinement    

     •  Mesh quality report

     •  Spatially varied  interpolation  

        methods including

           •  Triangular (TIN)

           •  Inverse Distance Weighting

           •  Nearest neighbor

           •  Sampling, and many more 

 

Hydraulic Components

     •  Culverts using FHI formulation

     •  Bridge pier drag forces   

     •  Weirs and Gates

     •  Internal hydraulic structures

     •  Sources and sinks

     •  Rainfall and evaporation

 

Boundary Conditions

     •  Water discharge hydrograph

     •  Water discharge and elevation 

        vs. time

     •  Water elevation vs. time

     •  Uniform flow 

     •  Free outflow 

 

Input Data Formats

     •  Metric or English units

     •  ASCII X, Y, Z   

     •  ESRI ASCII grid files

     •  USGS DEM

     •  ESRI Shape files    

     •  Autodesk DXF

     •  TIFF, GIF, BMP raster’s

 

Initial Conditions

     •  Dry-bed

     •  User-defined water elevations

Numerical Engine

     •  Spatial discretization using

        triangular elements

     •  High performance Finite-Volume

         engine 

      •  Double-precision computations 

        for higher accuracy

     •  Fully parallelized with OpenMP 

        for faster execution in Multiple-

        Core computers 

     •  >100X faster simulations 

 

Output Options

     •  Results at cross sections and 

        profiles

     •  Observation points

     •  Dynamic plots while the model 

        runs

     •  Maximum velocity and water 

        elevations

     •  Depth, velocity, bed elevation, 

        erosion or deposition, Froude

        number,  bed shear stress and

        sediment discharge            

     •  Polygon, line and point ESRI

        Shape files     

     •  Autodesk DXF mesh and 

        velocity fields

     •  GIS post processing plots 

        including shape files and raster 

        images

            •  Velocity field, depth and 

               WSE

            •  Bed shear stress

            •  Erosion and deposition

            •  HDF5

            •  VTK

 

Hardware requirements:

Hydronia recommends using one of the following cards to run RiverFlow2D Plus GPU

  • NVIDIA Tesla c2075
  • NVIDIA Geforce GTX 780

Note that the performance will highly depend on the card installed. Older generation NVIDIA cards may not allow the model to work properly. Use only one of the recommended cards above.

 

Hydronia has tested RiverFlow2D Plus GPU In the Tesla c2075 card that offers the highest performance gain so far in the order of 30x depending on the mesh size. The GTX 780 card is more economical and provides speed-ups of 12-15x depending on the number of cells.

We are now testing the new generation Tesla K20 that is expected to provide much higher performance than the previous cards. We will publish the results with this card as soon as the tests are completed.