OVERFLOW 2.5 introduces the ability to run OVERFLOW on GPU systems.
CPU Path
Only refactoring has occurred in the CPU path of the code. Users should see equivalent behavior as 2.4d. There might be a slight difference (round-off) in some of the force integrals due to a slightly different order of operations.
GPU Path
To run on the GPU set USE_GPU=.T. in the &GLOBAL namelist. Not all of the capabilities in OVERFLOW have been ported to run on the GPU. For the most part, the ported options will obtain either identical or very similar results on both the CPU and GPU. See doc/readmes/README.gpu.txt (duplicated below) for a complete list of supported options, instructions for running and compiling, and some other notes.
Build System
Makefile build system has been removed. Users must now use the CMake build system. See doc/readmes/README.cmake.txt for more information.
Added -DGPU cmake option to enable/disable building GPU path (default False)
Added -DSWAPENDIAN cmake option to enable/disable byte swapping in nvfortran
Starting with OVERFLOW 2.5.0, OVERFLOW supports running on NVIDIA GPUs (tested on V100, A100, H100, and GH200). This version (2.5) is the first providing this capability. As such, there are many options, described below, that are not supported yet. While we have done extensive internal testing, and some testing by external beta users, you may still encounter issues (primarily with building and running the code on your system). If issues are en …