-
Notifications
You must be signed in to change notification settings - Fork 1
Requirements
This project implements a GPU-accelerated hydrological simulation with MPI support. It is written in C++17 and uses CUDA for device-side computations. The build system is managed using GNU Make and requires several external libraries.
-
C++ Compiler Must support the C++17 standard. Recommended compilers:
-
g++(version 9 or later) -
clang++(version 10 or later) - Intel
icc(for optimized performance on Intel CPUs)
-
-
CUDA Toolkit (version 12.x or newer) Required to compile and run the CUDA kernels.
-
nvccmust be available in the system path. - Requires a CUDA-capable GPU with compute capability ≥ 9.0.
-
-
GNU Make Required to build the project using the provided
Makefile. -
MPI (Message Passing Interface) Required for multi-node and multi-GPU distributed simulations.
- Supported implementations:
OpenMPIorMPICH. -
mpicxxmust be available in the path.
- Supported implementations:
-
NetCDF (version 4.9.2) Required for reading forcing data and writing simulation outputs.
-
Must be compiled with HDF5 support.
-
The
NETCDF_PATHenvironment variable must point to the NetCDF installation directory, which must contain:-
include/— for headers files -
lib64/— for compiled libraries
-
-
-
HDF5 (version 1.14.4) Required as a dependency of NetCDF for high-performance I/O.
Getting Started
User Guide
Programmer Guide