Kokkos Kernels: Performance Portable Sparse/Dense Linear Algebra and Graph Kernels
Authors:
Sivasankaran Rajamanickam,
Seher Acer,
Luc Berger-Vergiat,
Vinh Dang,
Nathan Ellingwood,
Evan Harvey,
Brian Kelley,
Christian R. Trott,
Jeremiah Wilke,
Ichitaro Yamazaki
Abstract:
As hardware architectures are evolving in the push towards exascale, develo** Computational Science and Engineering (CSE) applications depend on performance portable approaches for sustainable software development. This paper describes one aspect of performance portability with respect to develo** a portable library of kernels that serve the needs of several CSE applications and software frame…
▽ More
As hardware architectures are evolving in the push towards exascale, develo** Computational Science and Engineering (CSE) applications depend on performance portable approaches for sustainable software development. This paper describes one aspect of performance portability with respect to develo** a portable library of kernels that serve the needs of several CSE applications and software frameworks. We describe Kokkos Kernels, a library of kernels for sparse linear algebra, dense linear algebra and graph kernels. We describe the design principles of such a library and demonstrate portable performance of the library using some selected kernels. Specifically, we demonstrate the performance of four sparse kernels, three dense batched kernels, two graph kernels and one team level algorithm.
△ Less
Submitted 22 March, 2021;
originally announced March 2021.
