This repo contains a starting template for developing and optimizing mixed-precision algorithms for IoT, ML/DL/RL, SDR, and DSP applications that can be deployed on custom hardware accelerators while at the same time having an efficient collaboration with the host CPU.
Once you use this template, you will need to adjust the status badges below to point to your repo.
| System | Status | More information |
|---|---|---|
| FOSSA Status | Open-source license dependency scanner (update URL after forking) | |
| GitHub Actions | Latest Linux/MacOS/Windows builds and regression tests |
This repo uses git submodules. The first step after pulling the repository is to configure the submodules:
> git submodule init && git submodule update
After that, the repo is ready to be build:
> mkdir build
> cd build
> cmake ..
> make
This will build the libraries, the CLI command projects, and the tests in test/mpadao_tests.
The solver examples use boost::multiprecision::cpp_bin_float_quad as a reference type for comparing numerical accuracy against Universal number types. Boost is optional — without it, the solvers are simply skipped.
Linux (Ubuntu/Debian):
sudo apt install libboost-devmacOS (Homebrew):
brew install boostWindows:
Download a Boost release from boost.org and extract it. Since we only use Boost's header-only multiprecision library, no build step is required. Point CMake to the source tree:
cmake -DBOOST_ROOT=C:/local/boost_1_86_0 ..CMake will automatically detect installed Boost on Linux and macOS. If Boost is not found, everything except the solvers directory will build normally.
When using VSCode, the repository contains a devcontainer spec in the directory $MPADAO_ROOT/.devcontainer.
The development container ships both GCC 14 and Clang 18 on Ubuntu 24.04 LTS:
# Build the container locally
docker build -t stillwater/mpadao:latest docker/
# Run interactively with the repo mounted
docker run --rm -it -v $(pwd):/home/dev/mpadao stillwater/mpadao:latestGCC 14 is the default compiler. To build with Clang instead:
CC=clang-18 CXX=clang++-18 cmake ..You can also build natively. The .gitignore of this repo filters out the following directories:
build/
build_msvc/
build_gcc/
build_clang/
You can use these build directories to organize your native and specific build containers so that they can run concurrently. For example, you can use the build/ directory to hold native builds, and build_gcc/ directory to hold the default build container builds.
To install the command line tools for ease of use, issue the install target:
> make installThis command will populate the $MPADAO_ROOT/bin, $MPADAO_ROOT/lib, and $MPADAO_ROOT/include directories, where $MPADAO_ROOT represents the directory path of the mpadao-template repository clone.
If you are on a Linux or MacOS system, you can add the bin directory to your path to pick up the command line tools:
> export PATH=$PATH:$MPADAO_ROOT/binFor Windows, use the environment variable editor to do the same.
To just build the projects in mpadao-template and ignore build targets in Universal and MTL5, use:
> cmake -DBUILD_DEMONSTRATION=OFF -DENABLE_TESTS=OFF ..To enable the Abseil logging example, add:
> cmake -DMPADAO_ENABLE_ABSEIL=ON ..If you want to update the submodules to the latest version of the upstream repos, issue this command:
> git submodule update --remote --mergeThe following figure shows the project structure of this repository:
After forking this template, update the following to match your project:
- Project name: Update
project()in rootCMakeLists.txt - CMake variable prefix: Rename
MPADAO_*variables if desired - Namespace: Replace
mpadao::insrc/lib/version/with your project namespace - Version: Set
MPADAO_MAJOR,MPADAO_MINOR,MPADAO_PATCHinCMakeLists.txt - GitHub URLs: Update badge links and
HOMEPAGE_URLinCMakeLists.txt - FOSSA badge: Point to your own FOSSA project
- Docker images: Update
stillwater/mpadao:*references in.devcontainer/anddocker/ - CI workflow: Update
.github/workflows/cmake.ymltriggers and branch names - README: Replace this checklist with your project description

