Home
GPU CUDA Guide
The EDEM CUDA GPU module uses the latest General Purpose Computing on Graphical Processing Units (GPU) technology.
GPU CUDA API Guide
CUDA requires that all functions be declared according to where they are going to be called and executed.
GPU CUDA Main Functions
The main function naming in the CUDA API follows a convention where the file name should not prefix the main function names.

Welcome
What's New
New features and enhancements available in EDEM.
Release Notes
This document contains information about the EDEM release, such as key features and enhancements, bug fixes, and known issues.
About EDEM
EDEM is a high-performance simulation software for bulk and granular materials.
Get Started with EDEM
The EDEM user interface comprises three icons for each simulation component - Creator, Simulator, and Analyst.
EDEM Creator
EDEM Creator is the pre-processor for creating an EDEM simulation model, including setting up equipment and defining the generation of bulk materials.
EDEM Simulator
EDEM Simulator is a powerful solver used for fast and efficient simulations.
EDEM Analyst
EDEM Analyst is the post-processor used to analyze and visualize the results of your simulation.
EDEM Contact Models
A Contact Model describes how elements behave when they come into contact with each other.
EDEM Cal Utility
EDEM Cal is a utility which works with EDEM and is designed to assist in performing calibration.
GPU CUDA Guide
The EDEM CUDA GPU module uses the latest General Purpose Computing on Graphical Processing Units (GPU) technology.
- Set up Hardware
  This section describes the tests necessary to run EDEM CUDA GPU such as GPU Device and Connectivity tests.
- Set GPU CUDA Simulator Parameters
  The EDEM GPU CUDA solver engine will automatically allocate the required memory and resources as the simulation runs.
- Run GPU from the Command Line
  To run EDEM GPU from the command line, specify the simulator engine with the flag “-E” followed by the corresponding engine number.
- GPU CUDA Known Issues and Limitations
  The EDEM 2022.3 GPU CUDA Solver Engine contains the following known issues.
- GPU CUDA Troubleshooting
  The following tips may help resolve any issues while using the EDEM GPU Solver Engine.
- GPU CUDA API Guide
  CUDA requires that all functions be declared according to where they are going to be called and executed.
  - GPU CUDA Class Declarations
    You can use multiple different particle types with the CUDA API. The CUDA API operates with a generalized set of classes and functions.
  - GPU CUDA Getter and Setter Functions
    A range of Getter and Setter functions have been implemented in order to maintain backward compatibility going forwards with the CUDA API,
  - Debug GPU CUDA
    EDEM displays the errors in a dialog box when your simulation is started. An additional method for debugging which allows you to obtain real-time information from the simulation is to run an attached simulation through the command line.
  - GPU CUDA Main Functions
    The main function naming in the CUDA API follows a convention where the file name should not prefix the main function names.
  - GPU CUDA Data Precision
    In the CUDA API, you can specify the precision with which you want the data to be handled. This allows the API scripts to leverage the speed benefits of the hybrid and single precision solvers.
  - CUDA Asserts
    The CUDA Solver has an additional option to enable asserts for enhanced error diagnostics.
  - GPU CUDA Header Files
    The GPU CUDA Header files are not required to develop a CUDA API plugin, but can be used for a better editing experience including syntax highlighting and Intellisense. They also allow API users to quickly get the most recent versions of the various CUDA functions that have been added in between releases. The Header files for the EDEM CUDA API are now located at: INSTALL_DIRECTORY\EDEM\src\Api\Cuda.
Programming Guide
The EDEM Programming Guide describes the EDEM API.
Coupling Interface Programming Guide
The Coupling Interface Programming Guide provides an overview of how to use the EDEM Coupling Interface to couple with a generic CFD code and/or Multi Body dynamics codes.
EDEMpy
EDEMpy is a Python library for accessing EDEM data from .h5 files (the underlying EDEM file format). EDEMpy does not require prior knowledge of advanced programming.
EDEM Solver Capabilities
EDEM contains four solver engines.
Appendix
This section provides additional information about how to estimate simulation time, supported EDEM File types, attribute definitions, and so on.
EDEM Terminology
Definitions of terms used in EDEM.
References and Bibliography
This section cites the contributions made by various authors and sources that have been used as reference material for EDEM, and the specifics of how they have been applied can be found in the individual contact model sections.

GPU CUDA Main Functions

The main function naming in the CUDA API follows a convention where the file name should not prefix the main function names.

Examples of main functions inputs

__device__
void calculateForce(NCudaApiTypesV1_0_0::CApiElement& element1,
NCudaApiTypesV1_0_0::CApiElement& element2,
NCudaApiTypesV1_0_0::CApiInteraction& interaction,
NCudaApiTypesV1_0_0::CApiContact& contactData,
NCudaApiTypesV1_0_0::CApiSimulation& simulationData,
NCudaApiTypesV1_0_0::CApiContactResults& contactResult)
{
}

Particle Body Force models must contain the following function:

__device__
void externalForce(NCudaApiTypesV1_0_0::CApiParticle& particle,
NCudaApiTypesV1_0_0::CApiSimulation& simulation,
NCudaApiTypesV1_0_0::CApiFieldData& fieldData,
NCudaApiTypesV1_0_0::CApiTotalForce& result){..}
{
}

In addition to these functions, the configForTimestep functions allow custom functionality to be modified at each Time Step. Even if they are not used, it is mandatory to include the following functions:

__device__
void configForTimeStepParticleProperty(NCudaApiTypesV1_0_0::CApiCustomPropertyData& particleProperties)
{
}
__device__
void configForTimeStepTriangleProperty(NCudaApiTypesV1_0_0::CApiCustomPropertyData& triangleProperties)
{
}

The configForTimeStepParticleProperty is used for both contact and external force models, so when it is used in a plugin that combines calculateForce and externalForce models, those models must have the same custom properties and custom property order, otherwise the wrong values can get modified.

The configForTimeStepTriangleProperty does not get the actual element custom property values, rather the difference between values between Time Steps (this is not the same as triangle contact deltas). This was done for easier mGPU implementation as the triangle custom property values are unused in any calculations and are intended for post-simulation analysis.