2.3 KiB
2.3 KiB
Project: SNPLIB
1. Overview
This project (SNPLIB) is a high-performance computing library for Quantitative Genetics developed in C++. The core objective of the project is to leverage modern C++ features and achieve extreme numerical computation performance through optimized math libraries (such as Intel MKL or OpenBLAS).
This project uses CMake for cross-platform build management and strictly adheres to the Google C++ Coding Style.
2. Core Tech Stack & Standards
2.1 Programming Language
- C++: The project will use C++ (C++17 or higher recommended) as the primary development language.
2.2 Build System
- CMake: This project uses CMake (version 3.15 or higher recommended) as the sole cross-platform build system.
CMakeLists.txtfiles are responsible for managing all compilation targets, dependency linking, and platform-specific configurations.
2.3 Performance & Dependencies
- BLAS/LAPACK Acceleration: To achieve high-performance matrix and vector operations, this project relies on the BLAS and LAPACK interfaces.
- Supported Libraries:
- Intel MKL (Math Kernel Library): The preferred high-performance library, especially on Intel architectures.
- OpenBLAS: A high-performance open-source alternative for BLAS/LAPACK with excellent cross-platform compatibility.
- The CMake configuration will provide options (e.g.,
USE_MKL=ONorUSE_OPENBLAS=ON) to allow users to select the math library to link against at compile time.
2.4 Coding Style
- Google C++ Coding Style: This project strictly adheres to the Google C++ Style Guide.
- Formatting: All C++ code submitted must be formatted using
clang-format(based on the Google style). - Linting: Using
clang-tidyis recommended to ensure code quality and style consistency.
3. Environment Setup
Before starting the build, you must ensure the following dependencies are installed on your system:
- C++ Compiler (e.g., GCC, Clang, MSVC)
- CMake (version >= 3.15)
- Intel MKL or OpenBLAS
- For Intel MKL, ensure the
MKLROOTenvironment variable is set, or point to its installation path via the CMakeCMAKE_PREFIX_PATH. - For OpenBLAS, ensure CMake can find it via
find_package(OpenBLAS).
- For Intel MKL, ensure the