Kokkos: Enabling manycore performance portability through polymorphic memory access patterns
By H. Edwards, Christian Robert Trott, Daniel Sunderland. J. Parallel Distributed Comput., 2014
947 citations recorded by the source corpus.
Subfields: Computer Networks and Communications, Hardware and Architecture
View Paper: https://www.semanticscholar.org/paper/602dcccc2bf6af1ca84355d530ff1e0a79391217
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Hardware and Architecture
Citations in the corpus
- The Athena++ Adaptive Mesh Refinement Framework: Design and Magnetohydrodynamic Solvers (2020, background): “…the mesh infrastructure from the physics allows each to be developed independently: for example a performance portable version of the AMR infrastructure based on the Kokkos library (Edwards et al. 2014) which can be run on heterogeneous architectures (including GPUs) is now under development.”
- Kokkos 3: Programming Model Extensions for the Exascale Era (2022, methodology): “As described in [9] Kokkos is a performance-portable programming model, implemented as a C++ library, that provides a number of abstractions to express both parallel execution and data structures.”
- Quantum ESPRESSO toward the exascale. (2020, background): “General-purpose frameworks have been proposed, 27–30 but none of them has reached maturity and widespread adoption.”
- The deal.II finite element library: Design, features, and insights (2019, methodology): “Multiple programming models are available for running code on a GPU, e.g., CUDA [53], Kokkos [54], RAJA [55, 56], OCCA [57], and OpenMP [58]. deal.”
- Learning local equivariant representations for large-scale atomistic dynamics (2022, methodology): “These operations are performed on the GPU by implementing an additional extension to LAMMPS’s KOKKOS package, which uses the Kokkos performance portability library [66] to entirely avoid expensive CPU work or CPU-GPU data transfer.”
- Retrieval-Augmented Generation for Scientific Code Understanding (2026, unknown): “Its use of MPI, Kokkos for performance-portable on-node parallelism [12], and heFFTe for distributed Fourier transforms [13] makes it a representative testbed for repository-level questions that combine software structure with high-performance-computing and numerical concepts.”