FEMaster

FEMaster is a powerful tool designed for solving linear finite element (FE) problems. It supports linear analysis and linear frequency extraction using high-performance algorithms and optional GPU acceleration. FEMaster offers various customization features, including support for multi-core processing via OpenMP, optional integration with the Math Kernel Library (MKL) for optimized performance, and CUDA for leveraging GPU computations. This tool is ideal for engineers and researchers looking to perform fast and efficient FE analysis on large-scale problems.

Table of Contents

• Features
• Installation
• Dependencies
• Optional Features
• Usage
  • Compiling the Project
  • Build Configuration
  • Compilation Options
• Testing
• Contributing
• License

Features

• Linear Analysis: Solve linear finite element problems efficiently.
• Linear Frequency Extraction: Perform frequency extraction for linear systems.
• CUDA Support: Optional GPU acceleration with CUDA.
• OpenMP: Enable multi-threaded execution for faster computations.
• MKL Integration: Use Intel's Math Kernel Library (MKL) for optimized performance.
• Cross-platform Support: Compatible with Linux, macOS, and Windows (including Apple Silicon support).

Installation

Step 1: Clone the Repository

git clone https://github.com/Luecx/FEMaster.git
cd FEMaster

Step 2: Install Required Dependencies

This project depends on several libraries such as Eigen (for matrix operations) and Spectra (for eigenvalue problems). It also makes use of argparse for command line arguments. The following script will download and install these libraries.

#!/bin/bash

# Create a directory to store libraries
sudo mkdir -p ~/libs
sudo mkdir -p /usr/local/include

# Install Eigen
cd ~/libs
git clone https://gitlab.com/libeigen/eigen.git
sudo cp -r eigen/Eigen /usr/local/include/

# Install Spectra
git clone https://github.com/yixuan/spectra.git
sudo cp -r spectra/include/Spectra /usr/local/include/

# install argparse
git clone https://github.com/p-ranav/argparse.git
sudo cp -r argparse/include/argparse /usr/local/include/

Save the above script as install_dependencies.sh, make it executable, and run it:

chmod +x install_dependencies.sh
./install_dependencies.sh

Step 3: (Optional) Install MKL for Optimal Performance

Intel's Math Kernel Library (MKL) is optional but provides optimized performance for numerical computations. To install MKL, follow the instructions on the official Intel MKL website.

Once installed, set the MKLROOT environment variable:

source /opt/intel/oneapi/setvars.sh

Step 4: (Optional) Install CUDA

For GPU acceleration, you can install the CUDA toolkit. Download and install it from the official NVIDIA CUDA website.

Dependencies

Required Dependencies

• Eigen: For matrix operations (download from Eigen website)
• Spectra: For eigenvalue computations (download from Spectra GitHub)

Optional Dependencies

• MKL: Intel's Math Kernel Library for optimized performance
• CUDA: NVIDIA's toolkit for GPU acceleration
• Google Test: For running the test suite

Optional Features

The project supports several optional features that can be enabled or disabled during compilation:

Feature	Description
OpenMP	Multi-threading support for parallel computation
MKL	Intel Math Kernel Library integration
Sequential MKL	Single-threaded MKL operation
CUDA Double Precision	High-precision GPU computation
Double Precision	Double-precision floating-point computation
Debug Mode	Enhanced debugging information

Usage

Compiling the Project

FEMaster provides several make targets for different build configurations:

make all      # Build everything (CPU, GPU, and tests)
make cpu      # Build CPU-only version
make gpu      # Build GPU-enabled version
make tests    # Build test suite
make clean    # Remove build artifacts
make info     # Display build configuration
make help     # Show available make targets

Build Configuration

You can view your current build configuration using:

make info

This will display information about:

• Project name
• Platform and architecture
• Compiler version
• Compiler flags
• Feature flags (MKL, OpenMP, Debug mode)

Compilation Options

The following build flags can be configured during compilation:

Flag	Default	Description
openmp	1	Enable/disable OpenMP support
mkl	0	Enable/disable MKL integration
mkl_sequential	0	Enable/disable Sequential MKL
cuda_dp	1	Enable/disable CUDA double precision
debug	0	Enable/disable debug mode
double_precision	1	Enable/disable double precision

To use these flags, append them to your make command:

make cpu openmp=0 mkl=1              # Build CPU version without OpenMP but with MKL
make gpu cuda_dp=1 double_precision=1 # Build GPU version with double precision
make all debug=1                     # Build everything in debug mode

Platform-Specific Features

The build system automatically detects and adapts to different platforms:

• Supports both x86_64 and ARM64 architectures
• Automatic detection of Clang vs GCC
• Platform-specific library paths (especially for Apple Silicon)
• Appropriate MKL configuration based on platform

Testing

The project uses Google Test for its test suite. To run the tests:

make tests
./bin/FEMaster_test

Test binaries are automatically configured for your platform, including proper library paths for both Intel and Apple Silicon machines.

Contributing

Contributions are welcome! Please submit issues and pull requests to the repository, and feel free to suggest improvements or new features.

License

FEMaster is licensed under the MIT License. See LICENSE for more details.