FLAME GPU 2.x
FLAME GPU 2 is currently in an pre-release (alpha) state, and although we hope there will not be significant changes to the API prior to a stable release there may be breaking changes as we fix issues, adjust the API and improve performance.
It is mostly feature complete with FLAME GPU 1 but may not meet the same levels of performance for all use cases. This is a result of the vastly more flexible interface used to template generated code. Performance is generally still very close to or better than FLAME GPU 1 and optimisations are always being added.
FLAME GPU 2 uses CMake as a cross-platform build system, with support for Windows via Visual Studio and for linux systems via make, with support for CUDA 11.0+. Please see the readme for further information on the dependencies of FLAME GPU 2 and the model development process.
To use FLAME GPU 2, you can either start by consider starting with one of the standalone models, or create a new standalone model based on the template repository or by downloading and building the main FLAME GPU 2 project and follow the instructions in the readme.
Pre-built python 3.6+ binary wheels are now available on some platforms. Please see the Release Notes for more information.
FLAME GPU 1.x
FLAME GPU 1.x releases are still available to download and use, although there are no plans for further development of FLAME GPU 1.
The latest release, FLAME GPU 1.5 supports CUDA 10.2 with Visual Studio 2015/2019 on windows and Make on Linux).
CUDA 11.0+ is unsupported.
The FLAME GPU 1 SDK requires an installation of a CUDA toolkit and compatible driver. The .zip download includes windows executables for each example model in the ‘bin’ folder for a x64 architecture. Visual studio 2015 project files are included for each of the SDK examples. These include custom build rules for FLAME GPU template processing and the CUDA build step (further details are provided in the documentation section of this site).