Kokkos: The Programming Model

C++ Performance Portability Programming Model

Kokkos Core implements a programming model in C++ for writing performance portable applications targeting all major HPC platforms. For that purpose 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 currently can use CUDA, HIP, SYCL, HPX, OpenMP and C++ threads as backend programming models with several other backends development.

The Kokkos Ecosystem includes:

Name	Info
kokkos	(this library) Programming Model - Parallel Execution and Memory Abstraction	GitHub link
kokkos-kernels	Sparse, dense, batched math kernels	GitHub link
kokkos-tools	Profiling and debugging tools	GitHub link
pykokkos	Provides Python bindings to the Kokkos performance portable parallel programming.	GitHub link
kokkos-remote-spaces	Shared memory semantics across multiple processes	GitHub link
kokkos-resilience	Resilience and Checkpointing Extensions for Kokkos	GitHub link

Related Work for the C++ standard library

Relevant and related efforts include:

Name	Info	Proposal
mdspan	Reference implementation of mdspan targeting C++23	P0009	GitHub link
stdBLAS	Reference Implementation for stdBLAS	P1673	GitHub link

Questions?

Find us on Slack: https://kokkosteam.slack.com or open an issue on github.

Website Content

• Quick Start
• Programming Guide
• Requirements
• Build, Install and Use
• CMake Keywords
• API: Core
• API: Containers
• API: Algorithms
• API: SIMD
• API in Alphabetical Order
• Deprecation for Kokkos-3.x
• Known issues
• Use Cases and Examples
• Kokkos Planning and Testing
• Tutorials
• Video lectures and slides
• GitHub Repo
• Contributing
• FAQ
• Citing Kokkos
• License