Kokkos on ThetaGPU
Kokkos implements a programming model in C++ for writing performance portable applications targeting all major HPC platforms. It provides abstractions for both parallel execution of code and data management. Kokkos is designed to target complex node architectures with N-level memory hierarchies and multiple types of execution resources. It can use OpenMP, etc as a backend programming model. For more information please visit https://github.com/kokkos/kokkos
The Kokkos shared memory programming model is a C++ library, that provides the necessary architecture specific backends (e.g. OpenMP, CUDA, …). To begin with, though, it is important to note that the Kokkos programming model is usable only in C/C++ codes. Hence, for those with Fortran codes, Kokkos must first be encapsulated within C/C++ functions and called from the main Fortran code.
The purpose of this document is to provide guidance on using Kokkos on Cooley. Please see the following pages for tutorials and more information on Kokkos: Kokkos GitHub and Kokkos Tutorials.
Using Kokkos at ALCF
ALCF provides assistance with build instructions, compiling executables, submitting jobs, and providing prebuilt binaries. For questions, contact us at email@example.com.
Building Kokkos on ThetaGPU
Please follow the steps below to build Kokks on ThetaGPU.
Step 1: Clone the repository
Step 2: Make sure that a relatively new version of CMake is available
Step 3: Create a build directory
Step 4: Generate the Makefile
Step 5: Install Kokkos
If you wish to install it in a different directory, change
CMAKE_INSTALL_PREFIX in step 4.